Find OpenCV Projects on Upwork with Vollna

Project: Fix and Enhance Existing OCR Readiness Assessment Platform

Hello,

I have an existing OCR Readiness Assessment Platform developed using Python, Streamlit, OpenCV and Tesseract OCR. Some features, especially image refinement and extracted-text accuracy, are not working correctly.

I need an experienced Python developer to fix and improve the existing application without breaking its current functionality. I will share the project repository after discussing the project with a suitable freelancer.

1. Project Overview

The platform evaluates document image quality before OCR processing. It calculates 10 quality factors, generates an overall OCR Readiness Score, extracts text using Tesseract OCR, and displays recommendations and analysis.

The 10 factors are:

1. Noise
2. Resolution
3. Blur
4. Contrast
5. Stroke Width
6. Text Density
7. Matra Continuity
8. Zone Integrity
9. Connected Component Stability (CC Stability)
10. Skew Penalty

The application already has image upload, quality-score cards, factor information, recommendations, OCR extraction, confidence highlighting, refinement previews and other dashboard features.

I want the existing application fixed and enhanced, not rebuilt from scratch.

2. Image Refinement for All 10 Factors

I need a properly working Refine button for all 10 factors wherever refinement is technically appropriate.

Refinement eligibility

- Score below 20: Disable refinement and recommend uploading a better-quality image or rescanning the document.
- Score from 20 to 80, inclusive: Allow refinement.
- Score above 80: Show that refinement is not needed.

These rules apply to individual factor scores, not the overall OCR Readiness Score.

Expected behavior when Refine is clicked

1. Identify the actual image-quality problem associated with the selected factor.
2. Apply a suitable image-processing technique for that specific problem.
3. Generate an actual refined image.
4. Recalculate the target factor and all other factor scores using the existing scoring pipeline.
5. Recalculate the overall OCR Readiness Score.
6. Show the original and refined images side by side.
7. Display before-and-after scores and validation results.

The refined image must show a meaningful visual difference related to the selected factor. Simply changing the score, label, recommendation or UI is unacceptable.

Examples of refinement targets:

- Noise: reduce unwanted noise.
- Resolution: apply appropriate enhancement without falsely claiming to recover missing detail.
- Blur: address blur using a technically suitable method.
- Contrast: improve inadequate contrast.
- Stroke Width: correct excessively thin or thick strokes where possible.
- Text Density: address density-related defects only when justified.
- Matra Continuity: preserve and repair appropriate Devanagari matras and Shirorekha.
- Zone Integrity: address actual structural defects in text regions.
- CC Stability: address unwanted noise components, fragmentation and irregular connected components.
- Skew Penalty: correct actual document skew.

The developer must inspect the existing implementations and select appropriate transformations. Do not blindly apply the same operation to every image.

3. Mandatory Safety Checks

Accept a refined candidate only if:

- The target factor improves by at least 0.5 points.
- No non-target factor decreases by more than 5 points.
- The overall OCR Readiness Score strictly increases.
- The actual image transformation is appropriate and does not unnecessarily damage the text.

If these conditions fail, reject the candidate and display a clear message such as "No safe refinement was found for this factor."

Do not manipulate scoring formulas or artificially increase scores to pass validation. If no safe improvement is technically possible, report that honestly.

4. Multiple-Factor Refinement

Add a feature allowing users to select multiple eligible factors and refine them in one workflow.

For example, users may select Blur, Contrast and Stroke Width, or CC Stability and Zone Integrity.

Requirements:

- Allow selection of multiple eligible factors.
- Apply the appropriate transformation for each selected factor.
- Process refinements in a controlled sequence.
- Use the currently accepted image as the input for the next refinement.
- Recalculate all factor scores and overall readiness after every accepted step.
- Apply the same safety checks to every step.
- Reject unsafe candidates without replacing the last accepted image.
- Display the original image, final accepted image and individual score changes.

If one refinement fails, report it clearly and allow other selected factors to be attempted where appropriate. Batch refinement must not bypass safety checks.

5. Fix Tesseract OCR and Extracted Text

The Extracted Text section sometimes displays incorrect or garbled text, incomplete words, unexpected characters and misplaced symbols.

Please inspect and fix the existing OCR pipeline, including:

- Tesseract configuration and language data.
- Hindi/Devanagari and English OCR support where appropriate.
- Image preprocessing, scaling, thresholding and page segmentation.
- Incorrect word segmentation and missing characters.
- Word-confidence values and confidence highlighting.
- Ensuring OCR processes the correct uploaded, selected or accepted refined image.

Extracted text must come from actual Tesseract processing. Do not hardcode or fabricate text. Test the results using suitable Hindi and English document images.

6. Refinement Preview and Controls

Provide:

- Original/current active image and refined candidate.
- Before-and-after target factor scores.
- Before-and-after overall OCR Readiness Scores.
- A table showing changes in all 10 factors.
- Safety-validation results.
- Apply, Cancel, Revert and Download functionality.

Apply should accept the candidate. Cancel should discard it. Revert should restore the appropriate previous/original image. Download should provide the accepted refined image.

7. Preserve Existing Functionality

Please inspect the codebase before modifying it. Do not unnecessarily rewrite or remove existing features, including:

- All 10 factor calculations, weights and scoring formulas.
- Overall OCR Readiness calculation.
- Image upload and region selection.
- Factor information and recommendations.
- OCR extraction and confidence highlighting.
- Visual diagnostics and overlays.
- Result history and storage.
- PDF report generation.
- Existing dashboard layout and styling.
- Working APIs, integrations and tests.

Make targeted changes. If a genuine scoring bug or architectural issue prevents correct behavior, explain it and implement the smallest safe correction.

8. Testing and Deliverables

Test the application on multiple document images, including Hindi/Devanagari images.

Verify:

- Eligibility rules for scores below 20, from 20 to 80, and above 80.
- Actual before-and-after visual improvement.
- Target factor improvement and non-target safety limits.
- Strict overall OCR Readiness improvement for accepted refinements.
- Individual and multiple-factor refinement.
- Apply, Cancel, Revert and Download.
- OCR extraction and confidence highlighting.
- Existing functionality through regression testing.

Deliver the updated source code, list of changed files, setup/run instructions and actual test results.

Do not claim a feature is complete unless it has been implemented and tested.

9. Freelancer Response

Please briefly describe your experience with Python, Streamlit, OpenCV, image processing and Tesseract OCR. Explain your proposed approach, estimated timeline and total cost.

Before final delivery, I want a working demonstration of actual image refinement, correct score validation, multiple-factor refinement, improved OCR output and preserved existing functionality.

I am looking for someone who can fix the underlying technical problems and deliver a properly working, tested solution—not just UI changes.

Skills: Python, Software Testing, Machine Learning (ML), OCR, Debugging, Image Processing, OpenCV, Computer Vision, Streamlit

My code ready main.py board.py eye.py brain.py. 5 blueprint photos. Need assemble to make LIVE EXE. Features Laptop Control Eye Vision Board BOS Order Block Hand Gesture Voice Hindi English Face ID. Tech Python OpenCV MediaPipe. Budget fixed $100 milestone wise. First check 2 files only. Indian dev preferred.

I make short educational videos about AI tools for business owners and sales teams. Most videos explain a new open-source project or AI tool.

I want a code-based editing pipeline that turns a script, a topic and a style reference into a finished vertical video with very little manual editing.

We already have a working base. You won't start from zero:
• A Remotion project that builds a video from a shot list, with 6 templates (split screen, UI demo, circle cam, flow tour, presenter behind, full frame)
• A QA script that checks script match, cuts, framing, lip sync, audio levels, clipped text and asset sources
• A HyperFrames test project
• A library of hook and pacing patterns, sorted by length and type

What's missing is the part that makes it repeatable. Today each video is assembled by hand, and scenes don't share a timeline. That's the job.


FIRST MILESTONE: AUTOMATED VIDEO ASSEMBLY

Build a pipeline that takes three inputs:
1. A topic (for example an open-source project page)
2. A style reference video that I provide as a file
3. A hook pattern chosen from our library

It must output:
• A finished 9:16 video, 30 to 45 seconds, exported at 1080x1920
• A side-by-side comparison with the style reference
• An editable project (Remotion or HyperFrames source) with all assets in one folder

The format we want:
• A short presenter clip (4 to 6 seconds) opens each video. It comes from an existing video API that we already use; you only call it.
• The rest is motion design plus screen recordings: the project page, its demo, captures with a visible cursor, and official logos used with permission or under the owner's brand guidelines.
• Large, readable title text on the first frame.
• A new visual beat every 5 to 8 seconds to keep pacing tight.
• Captions, a music bed under the voice, and sound effects only on visible events.

Must work this way:
• One shared timeline. Changing a scene's length moves the later scenes and their sound cues.
• One scene spec per video (JSON), built from reusable scene templates, not hand-coded scenes.
• Reference analysis. It measures the cuts, word timings, on-screen text and framing of the style reference and maps that pacing onto my voice track.
• Cheap checks first. The script and voiceover are checked before any paid API render.
• Runs from the command line on a Mac (Apple Silicon).


SECOND MILESTONE (AFTER THE FIRST IS ACCEPTED)

• Package the pipeline with clear docs and scripts so it can be run end to end from one command.
• Write a short runbook and record a walkthrough video.


TOOLS WE EXPECT

• Open source: Remotion and/or HyperFrames, ffmpeg, Whisper (mlx-whisper on Mac), Demucs, OpenCV or Apple Vision, Playwright for screen recordings
• APIs we already pay for: a video presenter API, a text-to-speech API, Cloudinary
• Languages: TypeScript (React) and Python


STYLE WE'RE AIMING FOR

Fast, clean educational explainers: a short presenter intro, then screen recordings and motion graphics with bold captions and tight pacing. I'll share specific style reference videos after we start.


HOW WE'LL ACCEPT THE FIRST MILESTONE

1. Five videos on five different topics, each built by one command with no hand edits after the run.
2. Every video passes our QA script, or you extend the script to cover the new format.
3. The presenter clip is 6 seconds or less per video, and API costs are logged per run.
4. The first-frame title is readable in 9:16, 3:4 and 1:1 crops.
5. Changing one scene's length in the spec re-renders correctly, with the sound cues moved.
6. Render time is under 10 minutes per video on an M-series Mac.


ABOUT YOU

• You've shipped Remotion or HyperFrames projects. Please share examples here on Upwork.
• You're comfortable with ffmpeg, audio timing and frame-accurate edits.
• You've used AI coding tools and can document a pipeline clearly.
• You write clear handoff docs.


TO APPLY, ANSWER THESE

1. Share one short video you built in code (Remotion, HyperFrames or similar) and say what you'd change about it.
2. How would you make scene changes move the sound effects automatically?
3. How would you check that a video's first-frame title is readable without watching it?
4. What would you build in your first 3 days?

## Project Overview

I am developing **HerdVision — an AI-Powered Livestock Intelligence Platform** focused on improving dairy farm monitoring in Pakistan using existing CCTV cameras and Artificial Intelligence.

I am looking for an experienced **Computer Vision / AI Developer** to help develop an initial working prototype, with the potential for long-term collaboration as the project grows.

## Project Requirements

The initial goal is to develop an AI system that can analyse dairy farm video footage and detect and track cows.

The developer should have experience with:

- Python, OpenCV, PyTorch or TensorFlow.
- YOLO or similar object detection and tracking models.
- Video analytics and CCTV/RTSP camera integration.
- Real-time video processing and AI model evaluation.
- Animal detection or behaviour monitoring (preferred).

## Initial Scope of Work

1. Assess the feasibility of using existing dairy farm CCTV footage.
2. Develop a prototype for cow detection and tracking.
3. Demonstrate the system on recorded video footage.
4. Provide a basic detection output with timestamps and visual indicators.
5. Deliver the source code, setup instructions, and technical documentation.
6. Recommend the next development steps for live CCTV integration and animal behaviour monitoring.

## Budget and Engagement

**Initial budget:** PKR 50,000–100,000, subject to the proposed scope and deliverables.

I prefer a fixed-price project with clearly defined milestones. Please suggest what can realistically be achieved within this budget, along with your estimated timeline.

## Ideal Developer

I am looking for someone who can demonstrate previous computer vision projects, explain the technical approach clearly, and build a reliable foundation for future development.

Experience in livestock monitoring, animal behaviour analysis, or agricultural AI will be an advantage.

**Important:** Please share relevant portfolio examples or working demonstrations, particularly involving video analytics, object tracking, or CCTV-based AI systems.

The long-term vision is to develop HerdVision into a commercial livestock intelligence platform capable of monitoring animal behaviour and helping farmers identify potential health and reproductive issues.

We are developing a digital-twin workflow for an oil & gas production facility using drone imagery, ground photography, video, LiDAR/RTK where available, and AI-assisted component identification.

We need a freelancer with experience in computer vision, 3D reconstruction/digital twins, Python, Power BI, photogrammetry, LiDAR, or oil & gas process equipment to evaluate the existing methodology and recommend/build the most practical workflow to:

reconstruct or organize the facility as a digital twin;
identify and count fugitive-emission components such as valves, flanges/connectors, open-ended lines, instruments, gauges, relief valves, pump/compressor seals, sight glasses, etc.;
assign each component a unique ID and avoid duplicate counting across multiple images;
associate components with process service such as Gas/Vapor, Light Oil, or Water/Light Oil;
create an auditable component database with image/location references and confidence level;
generate summary tables, component counts, QA/QC views, and reports using Power BI, Python, or another suitable platform;
recommend where existing AI models, object detection/segmentation, photogrammetry, or manual validation should be used.

Deliverable: a concise technical recommendation plus a working prototype using a sample facility dataset, showing the proposed digital-twin/component-count workflow and final reporting format.

Oil & gas production-facility experience is strongly preferred. Knowledge of YOLO/object detection, OpenCV, COLMAP/photogrammetry, point clouds/LiDAR, Python, Power BI, or GIS is a plus.

The objective is not simply object detection; it is to create a repeatable, auditable methodology for regulatory fugitive-component inventory and counting.

Scope is intentionally small; optimize for clarity and easy handoff in Python. For I/O heavy work, consider asyncio or multiprocessing only if it clearly helps—keep it simple first. The Python work centers on: OpenCV thumbnail generator. Turnaround is flexible; I care most about clean Python and reproducible setup steps.
Tech stack: Python 3 only. Reference batch ID: QA-PY-0107.

Skills: PHP, Python, Data Processing, Web Scraping, Software Architecture, Scripting, Data Analysis, Pandas

We are looking for an experienced Senior Software Developer for a long-term collaboration to build and maintain a highly intelligent desktop automation tool for PC applications/games.
​Key Responsibilities:
​Develop an automated workflow tool capable of reading UI elements, tracking task completion status, and performing human-like interactions.
​Implement Computer Vision (OCR / OpenCV / YOLO) to accurately read in-game/app text, menus, and completed objectives.
​Implement app lifecycle management (auto-launching, safe closing, and error handling/recovery).
​Apply human behavior simulation (randomized delay intervals, smooth mouse movement curves, natural click patterns) to prevent automated pattern detection.
​Maintain, update, and optimize the tool regularly to adapt to target software UI changes and updates.
​Technical Requirements:
​Advanced skills in Python (PyAutoGUI, OpenCV, EasyOCR/Tesseract, PyWin32) or C++ / C#.
​Strong experience with Process & Memory Management or Image Recognition-based UI automation.
​Experience in human-like interaction algorithms (Bezier curves, Gaussian distribution delays).
​Ability to build a clean GUI (PyQt/Electron/WPF) and secure licensing system for distribution.
​Note: This is a long-term position with ongoing maintenance and feature updates. Please provide examples of similar automation or computer vision tools you have developed in your proposal.

Skills: Python, Game Design, Software Development, Game Development, Data Analysis, Computer Vision, Automation, Game Testing, Game UI, Game Programming

🚀 Android Mobile Developer (Flutter/React Native + Real-time Computer Vision & Video Processing)
​Project Title: InTrainer Action Android App Development (Real-time Vision, Dynamic HUD & Media Generation)
​📌 Project Overview
​We are looking for an experienced Android Mobile Developer to build InTrainer Action, a performance fitness and exercise evaluation app. The app turns the smartphone camera into a real-time tracking system using computer vision (MediaPipe / OpenCV) and overlays a dynamic HUD to measure body movement, calories, and training pillars in real time.
​🎯 Key Features to Implement
​Real-time Camera & Pose Tracking (Long Duration):
​Continuous camera capture capable of running 1-hour workout sessions smoothly without phone overheating or frame drops.
​Integration with MediaPipe Pose / OpenCV for lightweight real-time joint and body landmark detection.
​Dynamic 5-Pillar & Calorie HUD Overlay:
​Live graphic overlay displayed over the camera preview showing cumulative burned calories and the InTrainer 5 Pillars: Strength, Stamina, Agility, Mobility, and Neural Control.
​Dynamic distribution of percentage values based on exercise biomechanics, always summing to a total of 100%.
​Simultaneous Video Capture & Smart Clip Generation:
​Automatic saving of the full workout video with the burnt-in HUD overlay straight to the gallery upon session completion.
​Instant post-workout automated clipping feature to export a 30-second "Best Moments" clip tailored for social media stories/reels.
​"Workout Certified" Certificate Generator:
​End-of-workout summary screen allowing users to input/toggle optional personal data (Full Name, Age, Height, Weight) and a profile or workout image.
​Custom Template Selector: Choice of visual themes/color schemes (Gold, Silver, Blue, Green, Purple).
​Generates a high-resolution story image showcasing the 5-pillar summary, total calories, a motivational quote, and a trainer's digital signature / verification badge confirming the workout was authentic and verified by AI.
​🛠️ Required Technical Skills
​Proven experience in Android development (Flutter, React Native, or Android Native / Kotlin).
​Hands-on experience with mobile computer vision frameworks (MediaPipe, OpenCV).
​Strong command over camera APIs (Camera2 / CameraX) and mobile video encoding/editing (MediaCodec, FFmpeg).
​Optimization expertise for real-time graphics rendering without excessive memory or battery drain.
​📝 What to include in your proposal:
​Examples of previous work involving camera feeds, real-time overlays, or computer vision.
​Links to portfolio apps published in the Play Store.
​Estimated timeline and fixed price/hourly estimate for the MVP release.

I am building an Android-only mobile application that puts intelligent video editing directly in the user’s pocket. The core engine must use on-device AI to:

• Automatically trim raw footage into meaningful segments
• Detect scene changes and tag each scene for quick navigation
• Offer a palette of real-time filters and effects that can be layered before export

A moderately detailed interface is ideal: clean timeline, quick-access buttons for the three smart actions above, and a preview window that updates instantly as the user scrolls through scenes or toggles effects. No crowded menus, yet enough visual cues to feel professional.

Please structure the code in Android Studio (Kotlin or Java) and keep inference on the device with frameworks such as TensorFlow Lite, PyTorch Mobile or a comparable library. OpenCV is welcome for any pre- or post-processing steps.

Deliverables
• Full source code, properly commented, compiling without errors
• Pre-trained models or training pipeline with clear instructions
• Signed APK (debug is fine) plus a brief screen-capture demo that proves trimming, scene tagging and filter application work smoothly on a mid-range device

Acceptance criteria
1. Trimming and scene detection each complete a 60-second 1080p clip in under 10 seconds on a Snapdragon 7-series device.
2. Accuracy: at least 90 % of scene boundaries correctly identified in a provided test set.
3. Filters apply in real time while scrubbing the timeline, with preview FPS ≥24.

Feel free to suggest additional optimizations, model compression techniques, or UX improvements that stay within the “moderately detailed” design approach.

Skills: Java, JavaScript, Mobile App Development, Android, Machine Learning (ML), Video Editing, Kotlin, OpenCV, Android App Development

I am developing a portable medical imaging device for seeing through the skin to visualize the location and depth of veins. I have successfully demonstrated the capability on a lab/bench top system with manual image processing performed after image acquisition. I need to automate the image processing functions so that I can demonstrate the capability in real-time.

Phase 1 of the work would be to provide the software for image acquisition, image enhancement, and display. Object segmentation for linear features representing veins is a possible addition.

Phase 2 would be hardware accelerated depth processing from stereo-pair images from two cameras. A user interface would be required for selecting two points for disparity calculation and calculating the difference in depth between the two points (vein depth under the skin).

I have an NVIDIA Jetson AGX Orin DevKit and compatible Allied Vision Alvium 1800 U-2050m cameras with USB interface.

A third phase of the software development would follow successful demonstration of the imaging capability using the Jetson hardware. Our plan is to port the software to an Android OS running on a Qualcomm chipset to miniaturize the final product into a smartphone-size device and/or AR-Glasses.

See attached examples of the images we have acquired and enhanced.

Stephen

# Custom YOLO Model and Dataset for The Division 1 (2016)

I'm looking for an experienced and affordable computer vision specialist to create a custom YOLO object detection dataset and trained model for **Tom Clancy's The Division 1 (2016)**.

## 1. Dataset Collection

- I will provide YouTube channels containing gameplay livestreams.
- The freelancer must extract approximately 3,000 unique frames containing visible enemies.
- The game has three main maps/environments relevant to this project.
- I will provide guidance to distinguish enemies from allies and friendly NPCs.
- Avoid duplicate or nearly identical frames.
- Include enemies at different distances, angles, lighting conditions and levels of visibility.

## 2. Image Annotation

- Annotate enemies with bounding boxes.
- Two classes: `Enemy_Body` and `Enemy_Head`.
- Annotate all clearly visible enemies in each image, not just one enemy per screenshot.
- Do not annotate friendly players or non-hostile NPCs as enemies.
- Provide the complete dataset in YOLO format, including images, labels and data.yaml.
- Please specify whether your pricing is per image or per bounding box.

## 3. Model Training

- Train a custom YOLO model, preferably YOLO11.
- Prioritize accuracy, low false positives and fast inference.
- Optimize detection of distant, small and partially visible enemies.
- Validate performance using gameplay footage not included in the training dataset.

## 4. Required Deliverables

- Approximately 3,000 annotated images.
- Complete YOLO dataset.
- Trained model: `best.pt`.
- Exported model: `best.onnx`, preferably FP16, 416×416.
- Training and validation results (mAP50, precision and recall).
- Short demonstration video showing the model detecting enemies.

## 5. Budget and Pricing

I'm looking for an affordable fixed-price solution.

Please provide two separate quotes:

**Option A — Dataset Only**

Extract 3,000 images from the provided YouTube gameplay videos, annotate enemies and heads, and deliver the complete YOLO dataset.

**Option B — Complete Project**

Everything in Option A, plus YOLO training, validation and delivery of the trained `.pt` and `.onnx` models.

Please include your total price, estimated delivery time, and examples of previous YOLO object detection projects.

I do not need a complete application or user interface.

Thank you!

We are looking for a specialist in facial recognition and biometric identity verification with hands-on experience with FaceTec (ZoOm / 3D Liveness) and AI-based methods for image recognition and deepfake protection.

The goal is to strengthen our identity verification flow against modern attacks: deepfakes, face swaps, injected video streams (virtual cameras / emulators), replay attacks, masks and other presentation attacks.

What you will do:
- Review our current face verification / liveness flow (FaceTec-based) and map its weak points
- Assess the real risks and known vulnerabilities of FaceTec and similar liveness SDKs against deepfakes and injection attacks
- Design and implement additional AI layers for deepfake detection and anti-spoofing (e.g. injection detection, frame/texture artifact analysis, device and camera integrity checks)
- Run adversarial tests: generate deepfake / face-swap / replay attempts and measure how the system responds
- Recommend thresholds, monitoring and fallback rules (manual review, step-up verification)

Required experience:
- Production experience with face recognition and liveness detection
- Hands-on experience integrating or operating FaceTec
- Practical experience protecting systems against deepfakes and presentation/injection attacks — real cases, not just theory
- Strong computer vision / deep learning background (Python, PyTorch or TensorFlow, OpenCV)
- Familiarity with ISO/IEC 30107-3 (PAD) and iBeta testing is a plus

Deliverables / definition of done:
1. Written vulnerability assessment of the current flow (attack vectors, likelihood, impact)
2. Test report with deepfake / spoofing / injection attempts and results (pass/fail rates)
3. Implemented or specified mitigation layer(s) with measured improvement (APCER/BPCER or equivalent metrics)
4. Recommendations document for thresholds, monitoring and manual review rules

In your proposal, please describe a concrete case where you protected a biometric system against deepfakes and what the outcome was.

Hi,

Need a quick Python developer to set up an automated OCR trading script on my Windows VPS for MetaTrader 5 (MT5).

Key Tasks:
1. Script Logic: Read a desktop screen bounding box (100, 100, 600, 400) using OpenCV / Tesseract OCR.
2. Signal Execution: Automatically execute a 0.01 lot order on XAUUSD via MT5 Python API when "BUY" or "SELL" text appears.
3. Single Trade Protection: Only 1 trade should be taken per signal. Avoid opening multiple/duplicate positions while the same signal remains on screen.
4. VPS Deployment: Connect via AnyDesk, set up the script, run a live test order, and verify.

Budget: $5 Fixed Price
Time required: 15 - 20 minutes for an experienced developer.

Quick 5-star review will be provided upon successful delivery!

I need a robust real-time face recognition solution that keeps working even when cameras are pointed straight into bright daylight. The system will be deployed both indoors and outdoors, so it must automatically adapt to sudden lighting changes as people move between shaded areas, hallways, and direct sun.

My main expectation is a working prototype that can:
• Detect and recognize faces with high accuracy under intense glare and shifting shadows.
• Run live on-site (edge device, GPU box, or similar) with latency low enough for surveillance use.

If you already have experience mitigating over-exposure, bloom, or lens flare with clever preprocessing, specialized sensors, HDR pipelines, or training tricks such as illumination-invariant embeddings, please let me know. OpenCV, TensorFlow, PyTorch, or any proven commercial engine is acceptable as long as the final stack can be handed over with clear build/run instructions.

I’ll provide access to camera streams for testing and a small labelled dataset; you are free to augment it as needed. Success will be measured by recognition precision/recall in direct sunlight during a live demo and the stability of the frame-rate under those conditions.

Skills: C Programming, Python, Matlab and Mathematica, C++ Programming, Face Recognition, OpenCV, Tensorflow, Computer Vision

Job highlights
8+ years experience in AI/GenAI with expertise in LLMs, RAG, Azure AI Foundry, LangChain, Agentic AI, Docker, Kubernetes

Design, build and deploy LLM, GenAI & Agentic AI solutions; lead multi-agent system architecture; implement enterprise-grade RAG solutions; optimize AI models; integrate with cloud and APIs; drive AI development using modern SDLC tools; ensure AI governance and ethics

8 - 13 Years

3 Vacancies

Salary: freelance monthly rolling contract (figures tbd)

Remote role

Must have key skills
AI Foundry,
Langchain,
LLM,
AIML,
Python

Other key skills
Tensorflow,
GenAI,
Artificial Intelligence,
Machine Learning,
Deep Learning,
Pytorch,
NLP,
Opencv,
Image Processing

Job description

What you’ll do
HIRING | 8yrs+ Lead AI Engineer GenAI & Agentic AI

Are you an experienced AI Engineer / GenAI Architect ready to build and lead next-generation LLM, RAG and Agentic AI solutions?

We are looking for Lead AI Engineers with 8+ years of experience, including strong hands-on experience in AI/GenAI, LLMs, RAG, Azure AI Foundry / Microsoft Foundry, and Agentic AI.

Experience: 8+ Years Role: Senior / Lead AI Engineer Domain: AI / GenAI / Agentic AI
Location: Remote role

Key Responsibilities
Design, build and deploy LLM, GenAI & Agentic AI solutions
Lead Multi-Agent System design and architecture
Architect and implement enterprise-grade RAG solutions
Leverage Azure AI Foundry / Microsoft Foundry for model orchestration, deployment, and evaluation
Optimize AI/GenAI/Agentic AI models for performance and scalability
Integrate AI solutions with Cloud, APIs and backend systems
Drive AI development using modern AI SDLC tools such as GitHub Copilot, Claude Code, Codex, etc.
Build and deploy solutions using CI/CD pipelines
Drive technical solutioning and collaborate with cross-functional teams
Contribute to AgentOps, AI governance, security, ethics and responsible AI

Mandatory Skills
8+ years of overall experience
Strong hands-on AI / GenAI experience
Azure AI Foundry / AI Foundry / Microsoft Foundry
Python
LangChain & LangGraph
LLMs & Generative AI
RAG Architecture
Agentic AI / Multi-Agent Systems
Docker & Kubernetes
CI/CD
Cloud & API integration
Enterprise AI Platforms

Experience with one or more:
Azure AI Foundry / Microsoft AI Foundry AWS Bedrock / AgentCore / AgentBricks Databricks Mosaic AI Gemini Enterprise Agent Platform / Agent Studio Snowflake Cortex Agents

Preferred Background
Candidates from Data Engineering, Data Science or Data-focused software engineering backgrounds are preferred.

2+ years of experience working on data projects will be an added advantage.

Soft Skills VERY IMPORTANT
Excellent communication skills are a must.

The ideal candidate should be able to:
Lead technical discussions • Communicate complex AI concepts clearly • Collaborate with engineering, data and business teams • Own end-to-end delivery • Mentor and guide technical teams

If you are an AI Engineer / GenAI Engineer / AI Architect / Lead AI Engineer with 8+ years of experience and strong expertise in RAG, LLMs, Azure AI Foundry & Agentic AI, we’d love to connect!

I’m putting together a project that centres on Artificial intelligence itself rather than a single sub-field, and I need a specialist who can take the lead in designing AI models from the ground up. The job involves shaping the architecture, writing clean, well-documented code, and guiding the overall model-building workflow so the end result is production-ready.

You will help me decide on the right algorithms, set up the training and validation loops, and implement any necessary preprocessing or feature-engineering that will give the model a real edge. While the exact domain—machine learning, natural language processing, or computer vision—may shift as the project evolves, the common thread is rigorous model design that meets clear performance targets.

Please be comfortable working in Python and familiar with libraries such as TensorFlow, PyTorch, or similar toolkits. If you prefer additional utilities like scikit-learn, Hugging Face, or OpenCV, feel free to suggest them; I value practical recommendations that speed up experimentation without compromising quality.

Deliverables will be:

• A documented model architecture (diagram or markdown is fine)
• Reproducible code with comments and environment specifications
• A brief performance report that explains metrics and next steps

If this sounds like the type of challenge you excel at, I’d love to hear how you would approach it and what timelines you foresee.

Skills: Python, CUDA, Machine Learning (ML), C++ Programming, Tensorflow, Pytorch, Hugging Face, AI Model Development

Hello,

I am researching weakly supervised video anomaly detection using the RTFM framework on the NWPU Campus dataset. I have tried several approaches but have not been able to reproduce the results reported in the TDSD paper.

I am looking for a Python/PyTorch research engineer who can help me:

1. Audit my existing RTFM implementation and experimental setup.
2. Check feature extraction, train/test splits, ground-truth alignment, temporal aggregation, and frame-level AUC evaluation.
3. Reproduce the relevant published baseline under the same evaluation protocol.
4. Identify the reasons for any performance gap and suggest scientifically valid improvements.
5. Provide reproducible code, documentation, and an explanation of the findings to support my own research paper.


Please also share your estimated cost for an initial debugging and reproduction task, and the expected deliverables.

Thank you.

Need a PyTorch image classification model for normal, smoke, and fire detection. The project includes preparing the dataset, building the model, and delivering a working solution for image classification. Experience with image-based classification tasks and model implementation is important. Please share relevant examples of similar work and your approach to dataset preparation and model training.

We are seeking an expert in AI object detection, specifically utilizing the YOLO (You Only Look Once) framework for analyzing satellite imagery. The ideal candidate will have extensive experience in training and optimizing YOLO models for high-resolution satellite data. You'll be responsible for developing accurate models that can detect and classify various objects in satellite images. Strong analytical skills and a solid understanding of image processing techniques are essential for this role. If you have a passion for AI and remote sensing, we'd love to hear from you!

The Project
I’m using a wide-angle USB camera mounted about 6 ft above a 12 ft × 6 ft area. The camera captures the whole area, but the image is warped. I need an accurate, rectangular, top-down image that supports reliable tracking of billiard balls.

What I Need

Evaluate camera options (the current camera plus any alternatives you recommend).
Correct lens distortion and perspective to produce an accurate rectangular image.
Ensure the result supports reliable tracking.
Build a repeatable solution that can be deployed at multiple locations without extensive manual calibration at each site.

Required Experience
OpenCV, camera calibration, lens distortion correction, and Python.

To Apply
Please describe your relevant experience, and tell me how you would approach calibration so it stays quick and repeatable at each new location. Also estimate the hours needed for an initial assessment.

We are looking for an experienced Computational Biology / Biomedical Image Analysis practitioner to provide hands-on technical mentorship to a junior computational biology student who is already comfortable with Python and wants to build practical capability in analysing biological image datasets.

The primary focus is to develop strong foundations in computer vision for biological and medical imaging, with particular interest in cancer-related datasets, microscopy images, histopathology, cellular imaging and other image-rich biological datasets.

This is not intended to be a passive tutoring engagement. We are looking for someone who can provide practical technical immersion using real datasets, guide the student through complete analysis workflows, explain the reasoning behind different methods and gradually transfer enough capability for the student to independently approach new image-analysis problems.

Expected Engagement

Approximately 1–2 hours per week

Initial engagement of at least 4 weeks

Remote, interactive and highly practical sessions

Opportunity to extend beyond 4 weeks depending on progress and fit

Key Learning Areas

The mentor should introduce and practically demonstrate the major stages of a biomedical image-analysis workflow, including:

Understanding biological and medical image formats and datasets

Image loading, visualisation and exploratory image analysis

Image preprocessing and quality enhancement

Resizing, normalisation, denoising and contrast enhancement

Thresholding and classical image-processing techniques

Feature extraction from biological images

Image segmentation and object detection

Cell, tissue or tumour-region identification

Morphological operations and contour analysis

Image classification

Quantitative extraction of biological features from images

Evaluation of image-analysis and computer-vision models

Understanding common pitfalls such as class imbalance, leakage and overfitting

Tools and Technologies

The student currently uses Python and would like practical exposure to tools such as:

OpenCV

scikit-image

NumPy / Pandas

Matplotlib

Jupyter notebooks

PyTorch / torchvision

TensorFlow/Keras where appropriate

Modern image-segmentation and classification approaches

MONAI or similar medical-imaging frameworks where relevant

Image annotation and visualisation tools such as Napari, QuPath or equivalent tools where appropriate

The mentor should help the student understand both classical computer vision and modern deep-learning approaches, including when each method is appropriate.

Practical Dataset Focus

Where possible, the sessions should use publicly available real-world datasets such as:

Histopathology / cancer image datasets

Microscopy datasets

Cell-image datasets

Tissue segmentation datasets

Medical imaging datasets

Other biological image datasets relevant to computational biology research

The student should progressively work through at least one complete mini-project from:

raw images → preprocessing → exploration → segmentation / feature extraction → modelling → evaluation → interpretation.

Suggested 4-Week Foundation

Week 1 – Image Analysis Foundations

Introduction to biological images, image representations, OpenCV/scikit-image, preprocessing, filtering, contrast enhancement, thresholding and exploratory image analysis.

Week 2 – Segmentation and Biological Feature Extraction

Hands-on segmentation of cells, tissues or tumour regions; contours, morphology, connected components and quantitative feature extraction.

Week 3 – Machine Learning and Deep Learning for Images

Image classification, CNN fundamentals, transfer learning, data augmentation and introduction to biomedical image models.

Week 4 – End-to-End Biomedical Image Project

Apply the techniques to a real biological or cancer dataset and complete an end-to-end workflow, including model evaluation, interpretation and discussion of possible research extensions.

The exact sequence can be adapted based on the mentor’s recommendations and the student’s progress.

What We Are Looking For

Ideal candidates should have practical experience in one or more of the following:

Computational biology

Bioinformatics

Biomedical image analysis

Computational pathology

Cancer imaging

Microscopy image analysis

Computer vision

Medical AI

Machine learning / deep learning applied to biological or medical images

Academic researchers, PhD candidates, postdoctoral researchers and practitioners working in biomedical AI or computational imaging are particularly welcome.

Important Mentoring Requirement

The objective is skill transfer, not task outsourcing.

The mentor should be comfortable:

coding interactively with the student;

explaining why a particular technique is being used;

providing small practical exercises between sessions;

reviewing the student’s code and approach;

introducing good research and coding practices;

helping the student learn how to independently explore unfamiliar image datasets;

connecting technical methods to biological interpretation.

By the end of the initial engagement, the student should have a solid enough foundation to independently begin working with biological image datasets and continue developing skills in computational imaging.

When Applying

Please briefly share:

Your experience with biomedical or biological image analysis.

Examples of relevant datasets or projects you have worked on.

The main tools you use for image analysis and computer vision.

Whether you have experience mentoring undergraduate or junior researchers.

How you would structure a practical 4-week learning programme for this engagement.

Your availability for 1–2 hours of live mentoring per week.

The priority is finding someone who combines strong technical competence, biological understanding and an ability to teach through practical work.

# AI Image Object Removal Engineer — Automated Pipeline for 5,000 Images/Day

**Job type:** Fixed-price initial prototype, with potential for ongoing work
**Experience level:** Expert
**Budget:** Please propose a cost-effective fixed price for the initial benchmark
**Ongoing processing budget:** Maximum $1 USD per day for compute

## Project overview

We are looking for an experienced AI/computer vision developer to build a fully automated image-processing pipeline capable of processing approximately **5,000 property photographs per day**.

The objective is to identify and remove specified unwanted visual elements from images while preserving the underlying photograph and maintaining high visual quality.

The system must operate automatically, process images in batches, handle failures reliably, and return validated results without requiring manual intervention for routine cases.

Cost efficiency is essential. We are particularly interested in lightweight models, efficient inference, batch processing and serverless GPU infrastructure such as RunPod, provided the complete solution meets our quality and cost requirements.

## Required functionality

### 1. Automated image processing

- Accept images through a batch-processing API or job queue.
- Support JPG/JPEG and AVIF images, including appropriate conversion where necessary.
- Identify target objects or unwanted visual elements.
- Generate accurate masks and remove the selected elements using AI inpainting or another suitable method.
- Preserve architectural details, textures, lighting, furniture, windows and other important image content.
- Return processed images in a specified output format.
- Preserve original images separately and never overwrite them.

### 2. High-volume processing

The target workload is approximately:

- 5,000 images per day.
- Approximately 150,000 images per 30-day month.
- Automatic processing without manual job submission.
- Configurable batch sizes and concurrency.
- Durable job tracking and automatic recovery after interruptions.
- Bounded retries, error reporting and quarantine of failed images.
- Idempotent processing to prevent unnecessary duplicate work.

### 3. Strict cost requirement

**The ongoing image-processing compute budget must not exceed $1 USD per day.**

This equates to approximately $0.0002 per image at the target volume.

RunPod Serverless, a dedicated GPU worker, lightweight CPU inference or another cost-effective approach may be proposed.

Applicants must provide a realistic cost model based on measured performance, not advertised GPU prices alone.

The estimate must include:

- Model initialisation and worker startup.
- Image decoding, preprocessing and encoding.
- Detection, masking and inpainting.
- GPU or CPU execution time.
- Retries and failed jobs.
- Idle time that is billed.
- Any other processing charges required to achieve the target.

Storage, data transfer and other infrastructure costs must be identified separately, with a complete estimate of total operating cost.

If the $1/day target is not technically achievable at the required quality, explain the cheapest viable alternative rather than promising unrealistic performance.

### 4. Image quality and validation

The system must not simply produce an output and assume it is correct.

Implement checks for:

- Successful image decoding and encoding.
- Correct association between input and output.
- Missing or incomplete outputs.
- Remaining unwanted elements.
- False-positive detection on images that require no modification.
- Unacceptable visual artefacts or damage to surrounding details.
- Output dimensions and format.
- Automatic rejection or quarantine of uncertain results.

Original files must remain recoverable.

### 5. Infrastructure and integration

The existing environment includes a production VPS running websites, databases, Python-based processing and background jobs.

The solution should minimise the resource footprint on the production server and preferably isolate intensive processing in a separate worker environment.

We are open to:

- RunPod Serverless.
- GPU containers.
- Efficient open-source computer vision models.
- Lightweight object detection and inpainting pipelines.
- Other suitable architectures that meet the quality, security and budget requirements.

Use version-pinned dependencies where practical, document model licensing and provide reproducible deployment instructions.

The system must not expose credentials, overwrite originals or require unrestricted access to the production server.

## Initial paid milestone: benchmark and proof of feasibility

Before building the complete system, the developer must produce a small working prototype and benchmark it on an agreed sample of authorised images.

The benchmark should measure:

1. Processing accuracy and visual quality.
2. Detection precision and recall.
3. Processing time per image, including startup overhead.
4. Actual compute cost per successfully processed image.
5. Projected daily throughput at 5,000 images.
6. RAM and GPU memory requirements.
7. Failure rate, retry overhead and recovery behaviour.

The benchmark must demonstrate whether the proposed system can meet the $1/day compute budget.

No large-scale deployment is required during this initial milestone.

## Deliverables

- Working benchmark prototype.
- Source code and setup instructions.
- Documented model and dependency versions.
- Measured throughput and processing cost.
- Image-quality evaluation and failure analysis.
- Recommended production architecture.
- Automated job queue and retry design.
- Security and operational considerations.
- Fixed-price proposal for completing the production pipeline.

All source code and custom work produced under the contract must be delivered to us, subject to the agreed contract terms and applicable third-party licences.

## Required experience

Please apply only if you have practical experience with several of the following:

- Python and computer vision.
- AI image inpainting and object removal.
- Object detection and segmentation.
- OpenCV, PyTorch or comparable frameworks.
- GPU inference optimisation.
- RunPod or other serverless GPU platforms.
- High-volume image-processing pipelines.
- Batch APIs, job queues and failure recovery.
- Measuring and reducing inference costs.

## Application questions

Please answer these questions in your proposal:

1. How would you process 5,000 images per day for no more than $1/day in compute?
2. Which model and infrastructure would you use, and why?
3. What measured inference speed have you achieved on a comparable workload?
4. How would you include model startup, retries and other billed time in your cost calculations?
5. How would you detect poor results and prevent incorrect images from being accepted?
6. What would your fixed price be for the initial benchmark?
7. What is your estimated price and timeline for the complete automated pipeline?
8. Can you provide a relevant example of a comparable image-processing system you have built?

**Important:** Generic proposals promising extremely low costs without benchmark evidence will not be considered. We prefer a small, measurable proof of feasibility before committing to a full implementation.

Are you a data-driven automation expert with strong experience in web scraping, product data processing, and bulk import automation? Do you excel at transforming complex supplier data into structured, import-ready formats? If so, we have an exciting opportunity for you!

About Us

FCI London is a luxury furniture and interiors company with a global team spanning the UK, India, the Philippines, Pakistan, South Africa, and the Middle East. We are seeking an E-Commerce Automation Specialist to fully automate product data ingestion, image processing, and pricing workflows for our website.

Your Role: Automate Product Management from Start to Finish

As an E-Commerce Automation Specialist, you will be responsible for building an end-to-end automated pipeline that extracts, transforms, and loads product data efficiently. Your core responsibilities include:

Web Scraping & Data Extraction:

Develop web scrapers to extract product data from supplier websites (descriptions, specifications, images, prices, etc.).
Ensure extracted data is structured, clean, and optimized for import into our system.
Automated Price List Processing:

Read and process supplier price lists from various formats (PDF, Excel, CSV, etc.).
Convert raw data into complex pricing combinations and structured CSV files that are ready for website import.
Implement dynamic pricing rules based on currency fluctuations, and supplier updates.
Image Processing & Optimization:

Automate image resizing, compression, and format conversions to meet website requirements.
Implement AI-based image enhancement techniques where applicable.
End-to-End Product Import Automation:

Develop scripts or tools to generate import-ready CSV files with all necessary product details.
Ensure seamless integration with our website.
Implement scheduled data updates to keep pricing, inventory, and product details current.
Automation & Workflow Optimization:

Identify repetitive tasks and create automation scripts to streamline product updates.
Experiment with AI tools and automation frameworks for better efficiency.
What We’re Looking For:

Technical Skills & Experience:

5+ years of experience in web scraping, automation, and e-commerce data processing.
Expertise in Python, Selenium, BeautifulSoup, Scrapy, or similar scraping frameworks.
Strong knowledge of Excel automation, CSV processing, and price calculations.
Experience with image processing tools (Photoshop automation, OpenCV, Pillow, or similar).
Understanding of bulk import/export processes.
Data Processing & Problem-Solving Abilities:

Ability to extract and transform unstructured data into clean, structured formats.
Experience handling complex pricing calculations and bulk product imports.
Strong knowledge of data validation, error handling, and automation best practices.
Process Automation Mindset:

Passion for eliminating manual work through automation.
Ability to build scalable, future-proof solutions for product and price management.
Experience with task scheduling and workflow automation tools.
Communication & Collaboration:

Strong problem-solving mindset and ability to work independently.
Excellent written and verbal communication skills.
Comfortable working in a remote, fast-paced, global environment.
Why Join Us?

Global Exposure – Work with a dynamic, multicultural team headquartered in London.
Remote Flexibility – Enjoy the freedom of working from home while making a tangible impact.
Impactful Role – Own and develop an automation system that revolutionizes product management.
Continuous Learning – Stay ahead with the latest web scraping, automation, and AI tools.
Who Should Apply?

Web scraping experts passionate about automating product ingestion and pricing.
Data engineers or automation specialists experienced in handling large-scale e-commerce data.
Tech-savvy professionals who love building smart, AI-powered automation solutions.
Problem solvers who can work autonomously and thrive in a fast-paced environment.
Apply Today!

If you're excited about automating the entire e-commerce product lifecycle, send us your CV and a short note on why you’d be the perfect fit. Let’s build a seamless, AI-driven product management system together!

Location: Remote (Work From Home)
Salary: Competitive, based on experience
Hours: Flexible to accommodate global collaboration
Job Type: Full-time
Job Type: Full-time

Benefits:
Work from home
Experience:

E-commerce: 6 years (Required)
total work: 6 years (Required)
Work Location: Remote

Skills: eCommerce, AI (Artificial Intelligence) HW/SW

I have a short segment of security-camera video showing two vehicles, recorded in completely dark conditions. The footage is low-clarity, so the license plates are currently unreadable.

Your task is to isolate the key frames, enhance them, and recover each plate number. Please supply both the cleaned-up still images (full frame and a cropped close-up of each plate) and a simple text file with the characters you were able to decipher. The work has to stand up to basic forensic scrutiny, so I need a clear methodology, minimal artifacts, and proof that the digits you provide genuinely originate from the source video.

You are free to use any combination of OpenCV, Topaz, AI super-resolution, de-blurring, or custom image processing—as long as the final output meets the goal: two verified plate numbers accompanied by evidence images that make their characters legible.

Skills: Image Processing, OpenCV, Video Processing, Computer Vision, Deep Learning, Image Recognition, Image Analysis, Data Annotation

ob Title:
Computer Vision / Biomechanical Engineer Needed for Markerless Sports Tracking Engine (Cricket/Sprinting)
Job Description:
We are building a markerless kinematic and kinetic performance tracking mobile application for fast bowling analysis, heavily inspired by platforms like Pitchwolf AI and Steffan Jones’ PaceLab frameworks.
We require an elite Computer Vision / Machine Learning Engineer with deep experience in pose estimation, object tracking, and structural biomechanics to build the core mathematical analysis engine. General app developers who only use standard off-the-shelf wrappers without deep math backgrounds will not be considered.
Key Engine Requirements:
1. Pose Estimation Overhaul: Implement YOLOv12-Pose, MediaPipe, or a custom network optimized for ultra-high-speed movement (240fps video) to eradicate keypoint jitter and tracking dropouts on fast-moving limbs.
2. Spatial Scaling & Calibration: Program an automatic pixel-to-meter translation mechanism calibrated via fixed real-world landmarks (e.g., cricket stumps or pitch markings).
3. Biomechanical Metrics Implementation: Code custom backend scripts using NumPy/SciPy to calculate:
• Thigh Angular Velocity (TAV): Dynamic change in femur angle over time, tracking rotational acceleration up to 7000°/sec.
• Stride Height Window: Dynamic vertical head-drop monitoring relative to standing baseline height (benchmarked against the elite 85-86% threshold).
• Ground Contact Time: Millisecond frame-counting algorithms separating BFC and FFC impact durations.
4. Inverse Dynamics / Force Fusion: Implement mathematical physics formulas (integrating Dr. Ken Clark’s sprinting mechanics frameworks) to indirectly compute ground reaction forces (GRF) and deceleration impact scores utilizing body mass and visual kinematics.
Technical Stack Preference:
• Backend: Python, OpenCV, MediaPipe, YOLO-Pose, NumPy, PyTorch.
• Deployment Environment: API-first architecture designed to feed a React Native / Flutter mobile frontend.
To Apply, You Must Answer This Question:
Briefly explain how you plan to handle severe motion blur on a cricket bowler’s arm during the release phase in a 240fps smartphone video clip, and how you will prevent tracking keypoints from jumping or collapsing. Applications that use automated generic templates or fail to answer this question directly will be deleted immediately.

Computer Vision & LiDAR Engineer Needed – Detect & Measure Pavement Uplift (3–4 Days)

Job Description

We are looking for an experienced Computer Vision / LiDAR Engineer to quickly develop and integrate a detection module into our existing mobile mapping system.

Our system is mounted on vehicles and uses LiDAR, RGB cameras and GPS/GNSS to collect data while driving through public spaces.

We already have a working data processing pipeline and an engineer who can guide you through the existing architecture.

What We Need

We want to automatically detect pavement uplift caused by tree roots (root heave) and measure its severity.

The solution should:

1. Detect root-related pavement uplift in sidewalks, cycle paths and paved public areas.
2. Measure vertical displacement (in cm) using LiDAR point cloud data.
3. Identify and locate affected areas using available GPS/GNSS data.
4. Combine RGB imagery and LiDAR data where useful to improve detection accuracy.
5. Integrate into our existing pipeline, so we can start collecting and processing new data immediately.

We will provide example images showing the types of pavement deformation we want to detect.

What We Already Have

* Working mobile mapping system with LiDAR and RGB cameras.
* Existing data collection and processing pipeline.
* Access to an engineer familiar with the system.
* Example images and available sensor data.
* Direct technical support during development.

You do not need to build a new platform or pipeline from scratch.

Timeline

3–4 days maximum. Immediate start required.

We need a working first version integrated into our system, allowing us to begin field testing.

Deliverables

* Working pavement uplift detection module.
* Estimated uplift height in centimetres.
* Georeferenced detections.
* Integration into our existing pipeline.
* Source code and brief technical documentation.
* Initial validation using available sample data.

Requirements

Experience with:

* Python and computer vision (OpenCV / PyTorch).
* LiDAR point clouds (Open3D / PCL).
* 3D geometry and surface deformation detection.
* Sensor data processing and integration.

Experience with mobile mapping or road surface defect detection is a strong advantage.

Budget & Contract

Fixed-price project.

Please include in your proposal:

1. Your relevant experience with LiDAR-based surface deformation detection.
2. How you would approach detecting and measuring pavement uplift.
3. Confirmation that you can start immediately and deliver within 3–4 days.
4. Your fixed-price quotation.

We are ready to hire immediately.

This is a hands-on engineering role focused on building reliable ROS / ROS2 systems in Python for production edge devices. You will work on device control, sensor coordination, motor control, camera pipelines, safety logic, node communication, health monitoring, and deployment reliability.
The ideal candidate is comfortable writing clean Python code for real hardware, debugging ROS nodes in the field, and making robotics systems stable enough for hospital environments.

Responsibilities

* Develop, maintain, and improve ROS / ROS2 nodes using Python.
* Build reliable communication between cameras, sensors, motors, lights, AI models, and cloud-connected services.
* Work on device logic for human detection, room activity, UV safety, sensor status, motor movement, and real-time state coordination.
* Implement ROS topics, services, actions, parameters, launch files, timers, and lifecycle behavior.
* Debug and improve node stability, timing, message flow, resource usage, and failure recovery.
* Integrate hardware components such as cameras, GPIO-controlled devices, motors, lights, relays, and sensors.
* Work with edge devices such as NVIDIA Jetson and Linux-based embedded systems.
* Build diagnostics and health checks for ROS nodes, hardware status, camera availability, sensor status, and system readiness.
* Improve logging, monitoring, restart logic, and fault recovery for production devices.
* Collaborate with computer vision, edge AI, cloud, hardware, and QA teams.
* Support field debugging and help analyze device behavior from logs, telemetry, and system state.
* Package and deploy ROS-based services using Docker or similar deployment workflows.

What We’re Looking For

* Strong Python programming experience.
* Hands-on experience with ROS or ROS2, preferably ROS2.
* Experience building real robotics or edge-device systems, not only simulations.
* Strong understanding of ROS nodes, topics, services, actions, parameters, launch files, and message types.
* Comfortable working with Linux, shell commands, system services, logs, and device-level debugging.
* Experience integrating hardware components such as cameras, sensors, motors, GPIO, serial devices, or microcontrollers.
* Ability to write robust, maintainable code for long-running production systems.
* Strong debugging skills across software, hardware, networking, and runtime issues.
* Understanding of real-time or near-real-time system behavior, timing, state machines, and failure handling.
* Ability to work independently in a fast-moving startup environment.

Nice to Have

* Experience with NVIDIA Jetson devices.
* Experience with Docker-based deployment for edge devices.
* Experience with camera pipelines, OpenCV, GStreamer, CSI/USB cameras, or video streaming.
* Experience with GPIO, I2C, UART, serial communication, relays, motors, or embedded control.
* Experience with ROS2 lifecycle nodes, diagnostics, launch systems, and multi-node orchestration.
* Experience with MQTT, cloud-connected IoT devices, OTA updates, telemetry, or fleet monitoring.
* Experience with robotics safety logic, human detection, or safety-critical automation.
* Experience working with computer vision models running on edge devices.
* Experience debugging field devices remotely.

Ideal Candidate

The ideal candidate is a practical robotics software engineer who can make ROS systems work reliably on real devices. You should be comfortable moving between Python code, ROS architecture, Linux debugging, hardware integration, and production reliability.
You should care about clean architecture, but also be willing to debug messy real-world issues such as camera failures, node crashes, timing problems, hardware disconnects, resource limits, and inconsistent field behavior.

Example Projects

* Build ROS2 nodes to coordinate camera input, AI inference, motor movement, and safety logic.
* Create a device state machine for room monitoring, human detection, UV readiness, and interruption handling.
* Improve restart and recovery logic when a camera, sensor, or ROS node fails.
* Build health-check nodes that report device status, node status, hardware status, and runtime errors.
* Integrate GPIO-controlled lights, motors, sensors, and relays into the ROS system.
* Improve launch files, configuration management, and deployment flow for multiple device versions.
* Debug production logs from deployed hospital devices and identify root causes.
* Package ROS services inside Docker for reliable deployment on NVIDIA Jetson.

I need a self-hosted identity-verification flow focused exclusively on U.S. driver’s licenses. The user journey is simple: they upload clear front-and-back images, your service analyses the images for authenticity (holograms, microprint, barcode consistency, etc.), extracts the data fields, and returns a verification status plus the parsed data in JSON.

The core task is to build this as a RESTful service and wire it into my existing backend. I’m comfortable exposing webhooks or consuming your API from Node.js, Python, or Java, so pick the stack you work fastest in—as long as the integration points stay standard (HTTPS + JSON).

Please rely on proven libraries or services for image analysis (e.g., OpenCV, AWS Rekognition, or any SDK you prefer) and keep everything packaged so it can run in our AWS environment.

Deliverables:
• Source code for the verification microservice
• API documentation (endpoints, request/response samples)
• Basic deployment script or Dockerfile
• A short README with setup instructions and any third-party licences

Acceptance criteria: A sample cURL call with an image returns a clear VERIFIED / NOT_VERIFIED flag, parsed name, DOB, licence number, and a confidence score of 0-100.

Timeframe is flexible but I’d like a first working demo within two weeks.

Skills: Java, Python, Amazon Web Services, Node.js, RESTful, OpenCV, API Development, Image Analysis

Need a freelancer to build an AI system that monitors vehicles being loaded onto a ferry and identifies the number of vehicles and each vehicle type. The system should detect trailer hitches, trailers, and supported vehicle types from a provided list. The project includes developing the full AI solution, helping specify camera and installation requirements, and integrating the system with our cloud-based ferry operations platform to deliver real-time loading data. Experience with computer vision and real-time data integration is important.

Please provide a budget that includes breakdown of software development, licenses, and hardware requirements to build and operate the system. We plan to have 2 locations that are monitored, preferably using the same AI system being fed from different sets of cameras.

You will recommend the tools, including programming languages and complete environment.

Skills: Java, JavaScript, Python, Software Architecture, Docker, OpenCV, Tensorflow, Computer Vision, AI Model Development, AI Integration

أحتاج إلى إعداد محتوى تسويقي مُحكم من 20 صفحة يشرح بتفصيلٍ رسمي عشرة طرق مجرَّبة للربح من الإنترنت للمبتدئين. أريد لهذا الدليل أن يركّز بالدرجة الأولى على التجارة الإلكترونية، مع إمكانية التطرق باختصار إلى مجالات داعمة مثل التدوين أو التسويق بالعمولة إذا خدم ذلك ترابط الأفكار.

أبحث عن كاتب لديه خبرة عملية في التجارة الإلكترونية وفهم واضح لسلوك المشتري العربي، وقادر على صياغة نص رسمي سلس يحرّض القارئ على اتخاذ خطوة عملية دون مبالغة ترويجية. يجب أن يتضمن كل محور:

• شرحًا مبسطًا للفكرة وخطوات البدء.
• أدوات أو منصات موثوقة يمكن الاعتماد عليها.
• مثالًا تطبيقيًا أو قصة نجاح مختصرة.
• نصائح عملية لتجنّب الأخطاء الشائعة.

تسليم العمل على ملف ‎.docx منسّق بالعناوين والفقرات، مع قائمة مراجع وروابط حيّة لأي أداة أو منصة مذكورة. سأراجع المحتوى على مرحلتين: مسودة أولى لملاحظاتي النهائية ثم النسخة المعتمدة. الدقة اللغوية والموثوقية أهم من الإطالة؛ كل كلمة ينبغي أن تخدم الهدف التسويقي وتُشعر القارئ بالاحتراف.

Skills: ASP, .NET, Android, Zen Cart, PSD2CMS, Amazon Kindle, Biztalk, OpenCV, Abnormal Psychology, Aerial Survey

We are looking for a **UK-based Computer Vision / Image Processing specialist** for a small, fixed-scope technical feasibility study.

The project involves converting complex images into simplified regions suitable for hand-embroidery patterns.

**Key requirements:**

* Strong experience with image segmentation and region detection
* OpenCV / Python image processing
* Colour clustering / quantisation
* Contour, polygon and geometry extraction
* Ability to preserve consistent region geometry across different outputs
* Experience with raster-to-vector or similar image simplification is beneficial

This is initially a **proof of concept**, not a full application build.

Please apply with **specific examples of comparable computer-vision/image-processing projects you personally worked on**, your role, and the techniques/technologies used.

**Important: You must be UK-based and personally carry out the technical work.**