Find OpenCV Projects on Upwork with Vollna

Looking for an AI/ML Engineer to build intelligent, data-driven solutions.
We need an expert with hands-on experience in Deep Learning, NLP, and Computer Vision. Projects may involve TensorFlow, Python, IoT integration, and AWS deployment.

If you have a strong background in machine learning and have built real-world applications using tools like OpenCV and Keras - let’s talk!

I’m looking for someone to write a data processing script for me using python and ray. I basically have a dataset containing 3d scene data comprised of room point clouds, furniture point clouds, furniture semantic labels and furniture position and orientation information.

The input for the script should just be the augmentation factor (If it is 5, each original scene is rotated by 5 different randomly chose angles and now dataset size is 6x original) and the path to the train and val datasets (ray / parquet).

The dataset has the schema:

Column Type
------ ----
vertices numpy.ndarray(shape=(1024, 3), dtype=float)
scene_id string
norms numpy.ndarray(shape=(1,), dtype=double)
centroid numpy.ndarray(shape=(1, 3), dtype=double)
semantic_class list
furniture_vertices numpy.ndarray(ndim=3, dtype=float)
positions numpy.ndarray(ndim=2, dtype=float)
rotations numpy.ndarray(ndim=2, dtype=float)
input_idxs numpy.ndarray(ndim=1, dtype=int32)
furniture_aabb numpy.ndarray(ndim=3, dtype=double)
room_centroid numpy.ndarray(shape=(1, 3), dtype=float)
room_scale numpy.ndarray(shape=(1,), dtype=float)

Each row contains an indoor scene.
vertices - room point cloud
scene_id - Identifier
norms - useful for denormalization
centroid - can be dropped
semantic_class - contains the semantic class of furnitures in the scene
furniture_vertices - contains the point clouds of furnitures in the scene
positions - contains target positions for furnitures in the scene
rotations - contains target rotations in euler degrees for furnitures in the scene
input_idxs - contains target input furniture slot index
furniture_aabb - can be dropped
room_centroid - centroid of the room vertices
room_scale - scaling factor

The dataset has a train and a val split.
It is a ray dataset in parquet format.
I have a visualization script that I can share which I will use to verify outputs.

The requirements are as follows:
1. Transform the dataset so that all the furniture in each scene is in canonical position (centered) and orientation. Adjust the rotation and orientation accordingly. Do this for both train and val splits.
2. Augment the train split so that all the scenes in their entirety are rotated about the vertical axis by random angles.

The delivery can include the scripts and I can run the processing on my end.

• Describe your recent experience with similar projects

🔍 What You'll Do
Design and build synthetic datasets of fake selfies and fake IDs using tools like:

Generative AI (GANs, FaceSwap, D-ID, Synthesia, Synthesis.ai)

Custom ID templates with OpenCV or graphics tools

Research and reproduce real-world fraud techniques:

We’re looking for an experienced AI developer or video automation engineer to help build a customizable video assembly system.

---

Project Overview:

We have a library of pre-recorded video clips (e.g., short instructional clips).

We need a system that can automatically generate personalized video sequences by combining different clips based on user inputs.

The system should add smooth transitions, optional voiceover prompts, on-screen timers, and basic interactivity cues.

---

Key Requirements:

Strong experience with video processing pipelines

Knowledge of ffmpeg, OpenCV, or other video editing frameworks

Experience with AI or machine learning for video/media automation (e.g., generative transitions, text-to-speech for prompts, or smart overlays)

Ability to automate assembly of modular videos into seamless, interactive sessions

Output must be playable in web/mobile formats (HTML5 video or streaming-friendly)

---

Deliverable (Phase 1):

A working prototype that assembles 3–5 short video clips into a personalized sequence, adds basic transitions, and outputs a file or playlist ready for web playback.

---

Why Work With Us?
This is an early-stage project with the potential to expand into an ongoing partnership. We're looking for someone who enjoys solving complex video processing problems and building automation tools that feel smooth and seamless to the user.

---

How to Apply:
Please share examples of similar projects you've worked on (e.g., automated video editing, AI-based media workflows, video pipelines). Briefly explain how you would approach stitching videos together dynamically with AI enhancements.

This project focuses on developing a computer vision module as a subcomponent of a larger system, specifically targeting the integration and configuration of multiple LiDAR cameras. The goal is to enable robust perception capabilities by fusing data from several LiDAR sensors, supporting advanced scene understanding and object detection.

I'm looking for a Python expert to help automate object detection using a video feed.

Key Requirements:
- Develop a Python script to perform object detection on a live video feed.
- Optimize the script for real-time processing.
- Ensure accuracy and reliability in various environments.

Ideal Skills:
- Proficiency in Python.
- Experience with computer vision libraries (e.g., OpenCV, TensorFlow).
- Knowledge in handling video streams.
- Background in automation tasks.

Skills: Python, Web Scraping, Software Architecture, Machine Learning (ML), OpenCV, Video Processing, Computer Vision, Object Detection

We are looking for an experienced proffesional will skills in CV and OCR to read & analyse videos.

As a small evaluation project will expect you to do this and submit ( we shall pay for this task too but an evaluation /test price)

Evaluation Task -

Record a video using screenrecoding and just click and show the flow of some working platform (any platform or Gmail). without any audio.. then see if the application will generate the right engaging voiceover script for that


Take a small video only but more than 2 mins and less than 5 mins

Check if any Opensource LLM/model can handle this? or use openAI .

Cutting the video at the right place (based on tracking the click function) and sharing that is important for the success of this task,


Share a working demo of this for discussion on the actual big project. ( 4 months min)

Hi there,

I've already built a React Native app with integrated React Vision Camera and Text OCR functionality. The UI is ready, and most of the work is complete. I now need your expertise to complete the final piece:

🔧 What’s Needed:
I want you to integrate automatic number plate detection and use OCR to extract and autofill:

Number Plate
State Name
This should work accurately for all US and Canadian license plates.

Existing Setup:
React Native app (screens are done)
GitHub Repo: https://github.com/ishitkaroli/camera
Tflite Pre Trained Model
React Vision Camera integrated
OCR is working

Your Task:
Implement a number plate bounding box detection (can use YOLOv8, TensorFlow, or TFLite compatible models)
Perform OCR only within the detected plate region
Autofill the extracted license plate number and state on the form
Optimize for speed and accuracy (real-time or near-real-time processing)

Reference:
I've attached a video demo showing exactly how I want it to work.

Platform:
React Native (iOS + Android)
ML/AI model can be either on-device (preferred) or lightweight cloud-based (only if needed)

Ideal Candidate:
Experience with Vision Camera and OCR
Background in mobile ML integration (TFLite/ONNX models)
Familiarity with US & Canada plate formats
Can work fast and communicate clearly
If you’ve done anything similar or know how to approach this efficiently, I’d love to hear from you! I need this work done in 2-3 days. Would be willing to pay extra if you have already done this feature before.

• What strategy would you follow to implement accurate number plate detection and OCR in this React Native app? Would you use an existing pre-trained model, train a custom one, or another approach?
• Have you worked on any number plate recognition or similar object detection + OCR projects before? Please share details or links.
• Can you implement this to run fully on-device?
• What’s your estimated timeline to complete this task?

I’m looking for a developer to build a voice-controlled GUI agent that works with RedBrickAI and OpenEMR to help radiologists perform hands-free tasks, improve workflow, and boost diagnostic efficiency.

Key Features:

Scroll through scan slices (axial, sagittal, coronal)

Adjust window settings (width, level, opacity)

Select tools (e.g., annotation, comment, zoom)

Execute predefined voice commands

Read and respond to scan interface elements (slice position, toggles, etc.)

RedBrickAI Task Examples:

1. Optimizing Visualization Settings
Workflow based on prompt: “Adjust window width to 171 and window level to 86”

Move to the relevant slider

Drag to the requested value

Adjust label opacity

Toggle “Visual pixel interpolation” on (blue), and others off (grey)

Reset to default when needed

2. Writing a Comment on a CXR or CT Scan
Workflow based on prompt: “Comment on left hilar opacity on the CXR”

Open annotation/comment tool

Click on target region (e.g., left hilum)

Enter comment: “There is an ill-defined opacity near the left hilum. Consider follow-up.”

Save and anchor it to the image

Goal:
To create a voice-activated agent that mimics a radiologist's manual workflow for faster, hands-free use of RedBrickAI or OpenEMR. Tasks include scrolling slices, adjusting contrast, enabling toggles, selecting tools, possibly even annotating, and reading out coordinates — all via voice input.

Ideal Skills & Experience:

Voice recognition (Whisper, Google Speech API, etc.)

GUI automation tools (PyAutoGUI, OpenCV, Selenium)

Experience with RedBrickAI or OpenEMR (preferred)

Strong Python and desktop/web automation background

Familiarity with medical imaging workflows is a plus

• Describe your recent experience with similar projects

🔹 PHASE 1: OCR IMAGE RECOGNITION MODULE
🎯 Objective:
Develop a tool to extract data from images or PDFs of circuit panel schedules, which can vary in structure and quality. The extracted data should be structured into a tabular format (like Excel) for easy editing and review.

✅ Key Features:
Image/PDF Upload Interface:

Drag and drop or file upload support

Accepts multiple file formats: JPG, PNG, PDF

OCR Engine:

Recognizes:

Panel Name

Load Description

Circuit Number (if available)

Amperage (optional)

Must handle a wide variety of layouts (e.g., horizontal/vertical text, inconsistent spacing)

Editable Table Output:

Each row = 1 circuit/load

Columns: Load Description | Panel (auto-filled or manual) | Notes

Supports inline edits and corrections by user

Optional Enhancements:

Highlight unrecognized or unclear text for user review

Save progress, export to Excel/CSV

🔹 PHASE 2: CIRCUIT SORTING & CATEGORIZATION TOOL
🎯 Objective:
Take the tabular circuit data (from Phase 1) and automatically sort and categorize it using built-in logic or AI assistance. Categories should be editable, with dropdowns and manual override.

✅ Key Features:
Auto-Categorization Logic:

Use keyword matching or AI model to classify based on Load Description

Example Categories & Sub-Categories:
Category Example Sub-Categories
Panel Sub-Panel
Lighting Washroom Lighting, Exit Lighting, Office Lighting, General Indoor, Stairwell Lighting, Emergency
HVAC Furnace, RTU, AHU, Circulation Pump, Baseboard Heater, Exhaust Fan
Plug-Load Coffee Pot, Microwave, Vending Machine, Copier, Server Room, Receptacles, Washer, Dryer
Other Disconnect, Fire Alarm, Transformer, Zamboni Charger
Mixed Use Not Sure
U/A U/A

Editable UI:

Dropdown selectors for Category and Sub-Category per row

Column filters and sorting

Highlight “Not Sure” rows for manual attention

Comment field per row (optional)

Export Final Results:

To Excel or CSV with columns:

Load Description | Panel | Category | Sub-Category | Notes

🧠 Notes & Considerations:
⚠️ OCR will need to handle unstructured formats and poor image quality — a major challenge.

⚠️ There must be manual override capabilities in both phases.

Web-based or lightweight desktop tool preferred.

Clean and user-friendly UI is essential.

💼 Ideal Candidate:
Strong experience with OCR (Tesseract, Google Vision, AWS Textract, etc.)

Knowledge of rule-based classification or AI/NLP

Frontend development experience (React, Angular, or Electron for desktop)

Experience with Excel/CSV output workflows

3D Try-On Technology Architecture
Given that the frontend is already completed, the focus will be on developing the backend and 3D functionality using open-source solutions to optimize for cost.
1. System Overview
This 3D try-on system will leverage open-source technologies to create 3D models, enable virtual try-ons, and integrate seamlessly with the existing e-commerce platform. The goal is to develop a lightweight yet effective system that provides users with an engaging shopping experience.
2. Key System Components
Frontend Integration
Since the frontend is completed, the integration of 3D models will be handled using WebGL and Three.js.
Technology Stack: Three.js (for 3D rendering), WebGL (for browser compatibility)
Key Features:
Rendering and manipulation of 3D models on the user's avatar.
Interaction with user input (rotating, scaling, zooming).
Support for real-time adjustments based on product choices.
3D Scanning System
The 3D scanning system allows for clothing to be scanned in real-time using video inputs and processed into a 3D model.
Technology Stack: OpenCV (for image processing), Meshroom (for photogrammetry)
Key Features:
Photogrammetry-based 3D model generation:
Use open-source software (e.g., Meshroom) to generate 3D models from multiple images.
Real-time scanning: Smartphone cameras, webcams, or professional cameras can be used.
Automatic feedback system to guide the user through the scanning process, ensuring optimal results.

Title:
Python Developer Needed for Desktop Workflow Automation and Database Integration

Description:
We are seeking a skilled Python developer to help automate data workflows and integrate custom scripts with a Windows-based office management platform. The goal is to streamline tasks such as note generation, appointment handling, and data entry, with all output ultimately being inputted into Management Software — either through its database, API, or user interface.

This project involves working with local software, custom Python logic, and secure, lightweight desktop automation tools.

🔍 Key Responsibilities
Create or refine Python scripts to:

Automate UI actions (clicking, typing, navigation)

Push/pull data from a MySQL database

Populate fields or forms inside Management Software

Build lightweight input interfaces (e.g., text prompts or summaries)

Ensure solutions are portable, efficient, and easy to run locally

Document setup and usage for team members

✅ Required Skills
Strong Python scripting experience

Proficiency in MySQL/MariaDB

Familiarity with:

PyAutoGUI, pywinauto, or similar for UI automation

requests, httpx for API integration

Tkinter, PySimpleGUI, or similar for basic GUI design

Experience interacting with Windows desktop software

💡 Preferred Qualifications

Image/template recognition tools (e.g., OpenCV, Tesseract)

Familiarity with speech-to-text or AI-powered note generation (e.g., Whisper, chatgpt)

Experience creating portable or standalone local apps for non-technical staff

📁 Deliverables May Include:
Python scripts that interact with the desktop platform and insert data into Management Software

GUI tools for summarizing voice input or user entry

Scheduling or appointment tools integrated into Management Software’s workflow

Documentation and instructions for setup and future updates

I understand you're looking for 5D hyperrealistic baby software that allows users to upload images (from the web or localhost) to generate detailed representations, similar to what "Baby Wonder" offers. This is a fascinating project with great potential.
To request a Machine Learning developer to design it, I suggest you provide them with the following key information:
1. Concept Description
"I am seeking a 5D baby visualization software, powered by Machine Learning, that enables expectant parents to obtain a hyperrealistic representation of their baby. The idea is for users to upload 2D or 3D ultrasound images, or even photographs of the parents, and the system generates a highly detailed, three-dimensional 5D image of the baby, showing facial features, movements, and expressions."
2. Key Features Requested
Detail the functionalities you expect from the software:
* Flexible Image Input:
* Support for uploading images from local files (localhost).
* Support for uploading images via a web interface (URL, direct upload).
* Ability to process different image formats (JPEG, PNG, DICOM for ultrasounds, etc.).
* Hyperrealistic 5D Generation:
* Creation of 3D baby models with realistic texture and detail.
* Simulation of movement and expressions (e.g., subtle smiles, blinking, hand/foot movements).
* Ability to adjust lighting and viewing angle for better visualization.
* Integration of a "5D" element, which could refer to time/movement simulation or the addition of sensations (though the latter would be more complex and perhaps outside the initial scope of an MVP). You might clarify if 5D refers to adding the dimension of time to 3D (i.e., a 3D video) or some other sensory feature.
* Personalization and Prediction:
* Ability to infer baby characteristics based on parent images (facial features, eye/hair color, etc.). This will require more advanced ML models.
* Option for users to adjust certain parameters (e.g., skin tone, hair type, if prediction is difficult).
* User Interface (UI/UX):
* Intuitive and easy-to-use interface for uploading images and viewing results.
* Functions to save or share visualizations.
* Underlying Technologies (For the Developer):
* Machine Learning/Deep Learning: This is fundamental. Mention the need to use generative networks (GANs) or other advanced models for image synthesis and feature prediction.
* Image Processing: Algorithms for segmentation, 3D reconstruction, and image quality enhancement.
* 3D Rendering: Use of libraries or frameworks for real-time or pre-rendered rendering.
* Web Platform: If a web-based solution is desired, a robust backend and an interactive frontend will be necessary.
3. Inspiration and Examples
Mention "Baby Wonder" as an example of what you're looking for, but highlight where you want to go beyond it or what features you find particularly appealing about that software.
4. Key Questions for the ML Developer
Be prepared to discuss the following with the developer:
* Technical Feasibility: How feasible is it to achieve 5D hyperrealism with current ML technologies?
* Dataset: What type and quantity of training data would be needed to develop a robust model (e.g., ultrasound images with associated 3D models, newborn baby images, etc.)? How could this dataset be acquired or generated?
* Recommended Technologies: Which ML frameworks and libraries (TensorFlow, PyTorch), image processing (OpenCV), and 3D rendering (Three.js, Unity, Blender) would they recommend?
* Time and Cost Estimates: A preliminary breakdown of project phases (research, model development, user interface, testing) and an estimate of necessary resources.
* Scalability: How the architecture can be designed to support a growing number of users if launched as a web service.
By presenting this information clearly and structured to a Machine Learning developer, you'll have a solid foundation to start the conversation and design this innovative software. It's an ambitious project with enormous potential.

Skills: Python, Machine Learning (ML), Face Recognition, OpenCV, YOLO

We have two 3D point clouds of the same haul truck—one empty, one loaded. Your tasks: (1) extract the truck‑bed geometry from each scan, (2) align/register both scans, and (3) calculate the total heaped load volume inside the bed, targeting ±1 % accuracy. Deliver within 2 weeks of data handoff. Final outputs: aligned LAS/LAZ files, volume report (CSV + PDF), QA visuals, and your reproducible scripts/source code with a brief algorithms list. NDA and secure file handling required.

Key Deliverables:

Aligned & trimmed point clouds (empty vs filled)

Load‑only surface mesh + empty‑bed reference

Volume report (CSV + PDF) ±1 % accuracy

Source code / scripts + algorithms list

Screenshots or short video QA walkthrough

• Describe your recent experience with similar projects
• Please list any certifications related to this project
• What techniques would you use to clean a data set?
• How do you deal with outliers or missing values in a dataset?
• What tools do you use for data mining and visualization?

I need a Python script that can generate stipple art (dot-based portrait) from an input image, using brightness values to control dot density.

🎯 Goal:
- Place more dots in *dark areas* (like hair), fewer in light areas (like face/skin)
- Dots must be *uniform in size* (0.9mm) — no variable dot sizes
- Dot placement should follow a *weighted Voronoi stippling* method (Lloyd’s relaxation or similar)
- The result must look balanced and *face must be clearly recognizable*
- Avoid equal dot density between hair and face

🖼 Input:
- PNG or JPG image
- Size: 1169 × 1190 px

📤 Output:
- PNG image with black dots on white background (no image behind)
- Or CSV file with (x, y) dot coordinates
- Adjustable: dot count (~15,000 -20,000) and image input

💻 Requirements:
- Pure Python script (preferred without GPU libraries)
- Efficient, clean code (usable on basic Windows laptop)
- Bonus: ability to preview result with optional GUI or save snapshots

📌 Reference:
I’m trying to achieve a similar result as shown here:
https://observablehq.com/@mbostock/voronoi-stippling

This is a good example of the output I’m aiming for:
[Attached: obama_stipple_reference.jpg]

💰 Budget: $18 – $24 USD
I'm open to negotiation depending on your experience and output quality.

📦 Files: I will provide a sample input image (sham.png) after contract starts.

If this sounds like your kind of project, I’d love to work with you!
1. Milestone 1: Basic Prototype (Preview with fixed image & dot placement)

Generate PNG with uniform dots

Density follows brightness

No background image — only dots on white
💵 Budget: $8



2. Milestone 2: Adjustable Parameters (dot count, input image)

Accept external images

Let user set number of dots

Optionally export to CSV
💵 Budget: $10-$12

Responsibilities:
Develop or fine-tune an AI/ML model to identify and remove backgrounds from product images.

Support batch image processing (hundreds of images at once).

Ensure high-quality output with crisp, clean edges around products.

Optionally, integrate a simple interface or script for internal use.

Optimize for speed and accuracy.

Requirements:
Strong experience with computer vision, deep learning, and image segmentation.

Familiarity with tools like TensorFlow, PyTorch, OpenCV, or similar.

Previous work with background removal or semantic segmentation is a big plus.

Ability to work independently and deliver production-ready code.

Good communication and documentation skills.

Preferred Qualifications:

Familiarity with U-2-Net, MODNet, or other background-removal models.

Ability to suggest improvements and best practices for photo quality and lighting.

Project Type:
One-time project with potential for follow-up support or enhancement.

Deliverables:
A trained model and/or script for batch background removal.

Documentation for running the process on our internal machines.

Sample results on a subset of our photos.


To Apply:
Please share:

Examples of similar projects you’ve worked on.

Which frameworks/tools you'd use for this task.

Estimated timeline for delivery.

We are seeking a reliable and proactive freelancer based in Japan to support a cutting-edge Vision AI project focused on real-world data collection. This work is critical to building large-scale training datasets for next-generation AI models.

This is a hands-on, field-based freelance role where AI meets the streets — ideal for individuals with a technical background who can manage edge systems independently and respond quickly to operational issues.

Responsibilities
* Oversee and monitor the Vision AI edge system daily (approx. 5–6 hours/day)
* Ensure uninterrupted and effective data collection
* Troubleshoot hardware and software issues in real time
* Maintain clear, simple logs and coordinate remotely with the technical team

Location:
On-site in Atsugi, Japan
(Applicants must currently reside in Japan)

Duration:
1 month (extendable to 2 months)

Start Date: Immediate

Who We’re Looking For
* Based in Japan and available for daily on-site work
* Background in engineering, robotics, computer science, or related fields
* Strong problem-solving skills and hands-on experience with hardware/software systems
* Reliable, detail-oriented, and able to work independently with remote guidance

Why Join Us?
* Work at the intersection of AI and real-world deployment
* Contribute to the development of next-gen computer vision datasets
* Collaborate with a global team working on advanced AI applications

If you're excited about contributing to meaningful AI innovation on the ground — we want to hear from you.

Looking for an experienced developer to build an MVP educational mobile app using Lovable.dev or similar AI-assisted coding platforms for cost-saving. After the MVP, full-time cooperation is possible once the MVP gains traction.

Key App Requirements:
- Full screen capture integration (MediaProjection API)
- Visper voice interaction implementation
- LLM integration (OpenAI or Gemini API)
- Voice generation/TTS (ElevenLabs, OpenAI TTS, or Google Cloud TTS)
- OCR text extraction (Google Cloud Vision or Tesseract)
- Mobile-optimized interface

Core Features:
- Real-time screen grabbing from any app/content
- Voice-to-text student queries
- AI-powered educational assistance with voice responses
- OCR processing of captured screen content
- Task-solving guidance

Possible Technical Stack:
- Frontend: React Native or Flutter
- APIs: Claude, voice TTS, OCR services
- Backend: Firebase or Supabase
- Audio: WebRTC or AudioKit
- Image Processing: OpenCV for preprocessing

Important Note: This MVP will be used as a sideloaded Android app (not through Play Store), allowing full screen capture functionality without store restrictions.

Ideal Candidate:
- Experience with Lovable.dev or similar vibe-coding tools
- Android development background (MediaProjection API experience preferred)
- API integration skills (LLM, TTS, OCR, voice processing)
- Comfortable with sideloading deployment requirements
- Ed-tech domain knowledge preferred

Timeline: 2-4 weeks for functional MVP
Budget: $1,500-$2,000

Please share portfolio examples of Android apps with screen capture functionality and your experience with rapid prototyping platforms.

Position Senior Full Stack Engineer with Relevant experience with AI Computer Vision.
Must to have 5 years of totaly experience
Preference for profiles between 5 and 10 years of total experience
Must to have Senior experience with React
Must to have Senior experience with Node.js
Must to have Senior experience with Nest.js
Must to have relevant experience with Fabric.js
Must to have relevant experience with TypeScript
Must to have relevant experience with Python
Must to have relevant experience with PostgreSQL
Must to have relevant experience with AWS S3 / Python-based AWS Lambda
Must to have relevant experience with Computer vision with Python
Must to have relevant experience with Computer vision with geometry, canvas transformations, image recognition of engineering drawings
Must to have relevant experience with mathematical concepts and coordinate systems
Must to have relevant experience with OpenCV
Must to have relevant experience with ClaudeAI
Must to have relevant experience with GitHub


Good level of english and good communication
CET Time zone
100 % remote
ASAP
Long term period
Must to have included the Github profile

Project
Functioning SaaS product in the manufacturing industry,
Opportunity to get hands-on experience with AI in the computer vision field, having the most variate challenges
Technologies, Programming Languages, and Frameworks:
The project utilized TypeScript and Python across different layers.
The frontend was built using ReactJS, Typescript, with Material UI for component styling and FabricJS for advanced canvas interactions.
The backend APIs were developed using NestJS with TypeScript and PostgreSQL for structured data storage.
For AI and image recognition tasks, we used Python-based AWS Lambda functions.
Tech Stack Overview:
- Frontend: ReactJS + Material UI + FabricJS (for dynamic drawing, annotations, and dimension editing)
- Backend: NestJS (TypeScript), PostgreSQL- AI/Computer Vision Layer: Python-based AWS Lambda functions- Infrastructure: AWS-native (S3 for file storage, Lambda for computer
AI & Computer Vision Approaches:
The AI component involves image recognition of engineering drawings and extracting dimension tolerances from visual data.
A combination of OpenCV for image preprocessing and prompt-driven inference (via ClaudeAI) for interpreting annotations.
The architecture supports both automated and manual workflows for annotation verification.
AI/ML Libraries & Tools:
OpenCV for core image processing tasks, and in some flows, the ClaudeAI API was leveraged to extract semantic meaning from structured or semi-structured drawing annotations
RESPONSABILITIES
Designing a high-precision, user-friendly canvas tool with FabricJS
Building a modular, scalable backend with NestJS and clean architecture principles
Working with cross-disciplinary challenges at the intersection of mechanical engineering and AI
Delivering a tightly integrated workflow between manual annotation and AI-assisted recognition
The codebase is large and deeply interconnected, with individual frontend and backend files running into thousands of lines and covering numerous conditions and edge cases. Needs to add task boards or written acceptance criterias,
The project involves advanced geometry, canvas transformations, and image recognition, which required a solid grasp of mathematical concepts and coordinate systems.
on a computer vision-heavy project at this level must to have interpret what the code was doing, particularly in areas involving tolerance extraction, bounding boxes, and annotation mapping.

PLEASE attach complete matrix and your cv to your application, thank you!!

We are looking for a skilled Android developer (Kotlin) to integrate a 3D face scanning feature into our existing mobile application.

Project Details:
-The app is already developed in Kotlin
-You’ll need to integrate a 3D face scan module (open to suggestions like OpenCV, ARCore, FaceMesh, or other libraries/APIs you recommend)
-The goal is to capture and render a 3D model of the user's face
- Clean UI integration and seamless performance are important

Deliverables:
-Integrated and functional 3D face scanning module
-Real-time preview and 3D mesh rendering
-Basic error handling and fallback mechanism
-Source code

Please include in your proposal:
-Past projects involving 3D face or object scanning (with links or screenshots if possible)
-Your approach or preferred library to achieve this
-Estimated timeline and budget

• Include a link to your GitHub profile and/or website

Job Posting: Robotics Engineer for Educational AI Robot

Project Overview

We are seeking a skilled Robotics Engineer to design and develop a simple, user-friendly, and safe educational robot for interactive learning. The robot should engage users in dynamic, fun, and effective learning experiences across a broad age range. It will incorporate AI (e.g., ChatGPT , gemini and many such via API) for voice interaction and educational content delivery, running on affordable hardware .

Key Requirements

Develop a Simple Educational Robot: Build a minimal, cost-effective robot focused on interactive learning, with basic movement (e.g., wheels screens), voice interaction, and educational features.

User-Friendly and Safe: Ensure the robot is intuitive to use and safe for users of all ages (e.g., rounded edges, low-power components, and simple controls).

Incorporate Engagement Features: Add interactive elements like images on screen using AI (e.g., AI API for natural language processing).

Adaptable Design: Create a robot that can be used in various learning environments (e.g., classrooms, homes, or workshops) with modular or customizable features.

API Integration: Integrate AI APIs (e.g., ChatGPT) for real-time voice interaction and educational content delivery.

Responsibilities

Design and prototype a simple robot chassis using CAD software (e.g., Fusion 360 or SolidWorks).

Program a Raspberry Pi (or similar microcontroller) for robot control, movement, and AI integration.

Integrate sensors (e.g., ultrasonic or basic camera) for basic environmental interaction or navigation.

Implement voice input/output using microphones and speakers, connected to AI APIs for interactive learning.

Ensure safety features (e.g., emergency stop, durable materials) and user-friendly controls (e.g., mobile app or simple buttons).


Test and iterate the robot to ensure reliability and educational value in diverse settings.

Ideal Skills and Experience

Robotics Engineering:

Proficiency in CAD design for simple robot chassis (e.g., Fusion 360, SolidWorks).

Experience with motor control (e.g., DC motors or servos) for basic movement.

Knowledge of sensor integration (e.g., ultrasonic sensors, good cameras) for environmental awareness.

Embedded Systems:

Expertise in programming Raspberry Pi or any microcontrolar using a any coding language

Familiarity with Robot Operating System (ROS) or similar frameworks for robot control.

Experience with GPIO programming for sensor and actuator integration.

API Integration:

Strong experience with ChatGPT API or similar NLP APIs for voice-based interaction.

Ability to integrate and optimize AI models for real-time educational content delivery.

Educational Technology:

Experience designing interactive learning tools or edtech products.

Understanding of interactive learning principles (e.g., gamification, adaptive learning).

Safety and User Experience:

Skills in designing safe, durable, and user-friendly robotic systems for diverse age groups.

Familiarity with UX principles for intuitive controls (e.g., mobile app or voice commands).

Additional Skills (preferred):

Knowledge of lightweight AI model deployment (e.g., LLaMA) on edge devices like Raspberry Pi.

Experience with audio processing for voice input/output.

Basic computer vision (e.g., OpenCV) for visual engagement features (if applicable).

Preferred Qualifications

Proven experience building autonomous or educational robots (e.g., Raspberry Pi-based projects).

Portfolio showcasing robotics or edtech projects, ideally with AI integration.

Familiarity with safety standards for educational products (e.g., CE or UL compliance).

Strong communication skills to collaborate on project goals and provide updates.

Application Process

Please submit:


A resume or portfolio highlighting relevant robotics, edtech, or AI integration projects.



Examples of similar projects (e.g., autonomous robots, voice-controlled systems).



A brief explanation of your approach to building a simple, safe, and engaging educational robot.


Familiarity with the referenced YouTube video (https://www.youtube.com/watch?v=e-nbSGRFP4Q&t=15s) is a plus.

Budget and Timeline

Budget: Flexible, based on experience (250k-500k inr).

Timeline: Prototype within 6 months, with iterative testing.

We’re excited to collaborate with a passionate Robotics Engineer to create an innovative educational tool that makes learning fun and effective!

Skills: Mechanical Engineering, Robotics, Software Development, Artificial Intelligence, Embedded Systems, Robotic Process Automation, Prototyping, Mechanical Design, Robot Operating System (ROS), AI Development

We are seeking a skilled developer to implement a Lens Distortion Algorithm based on provided blog and repository resources. The ideal candidate will have a strong background in both C++ and Python, with the ability to understand and translate complex algorithms effectively. Your expertise will be crucial in ensuring that the Python implementation maintains the accuracy of the original C++ version. If you have experience with image processing and algorithms, we would love to hear from you!

Why from C++ to Python? We want this as a basis to experiment on further, introducing own assumptions and tweaking later.

Blog: https://hh409.user.srcf.net/blog/lens-distortion-correction-lines.html
Repo: https://github.com/hugohadfield/LensDistortionFromLines

If you have worked with similar algorithms like this, that work well on real world images, we are also happy to go for a different approach.

---

Additional context to narrow down the scope even further. We need in python:
- Hough lines algo + iterative optimization algo
- Division distortion model (not their polynomial one)
- Canny edge detection (you can try to use OpenCVs canny edge) + their cleaning algo after the canny edge itself

I need a working Jupyter notebook, that can transform some provided test images using the new Python implementation.

We are seeking an experienced computer vision and data analysis expert to develop a script that automates the detection of freezing events for an array of water droplets from video recordings. The goal is to determine the precise temperature at which each individual droplet freezes.

We have video footage of an array of small droplets on a surface that is being cooled. As the temperature drops, the droplets freeze, causing a distinct change in their appearance from transparent to opaque. We also have a corresponding data log file that records the temperature at specific video frame intervals.

Project Description & Workflow:

The required script should perform the following sequence of tasks:

Droplet Detection & Tracking: The code must first identify and locate each individual droplet in the video frames. The droplets are arranged in a grid-like pattern, which should simplify detection. It should be able to track each droplet throughout the video.

Freezing Event Detection: For each droplet, the script needs to monitor its appearance frame by frame. The key indicator of freezing is a significant change in opacity. The droplet will transition from being clear (showing the metallic surface underneath) to a dark, opaque solid. The code must accurately detect the exact frame in which this transition occurs for each droplet.

Data Correlation: Once the freezing frame for a droplet is identified, the script must look up the corresponding timestamp or frame number in our provided data log file (e.g., a CSV file).

Temperature Extraction: From the data log, the script will extract the temperature associated with that specific frame.

Output: The final output should be a structured data file (e.g., CSV) that lists each droplet (e.g., by its coordinates or an assigned ID) and its corresponding freezing temperature.


Relevant Skills:
- Computer Vision
- Image Processing
- Algorithm Development
- Video Analysis
- Programming (Python, OpenCV, etc.)
- Data Analysis

We’re looking for a skilled AI Engineer with experience in generative image models and interactive frontend development to build the AI component of a creative design tool. You’ll be responsible for developing a system that allows users to input structured prompts and receive editable design outputs, which are then adapted to specific templates for production use.

This role is ideal for someone who thrives at the intersection of AI, design, and UX, and wants to bring practical generative tools to real-world creative workflows.

Key Responsibilities:
- Build a user-facing interface (Streamlit or similar) that allows users to:
- Input prompts and brand data (e.g. colors, slogans, style).
- Select from AI-generated design concepts.
- Make simple adjustments (text, image placement, cropping).
- Use OpenAI’s DALL·E 3 API or equivalent for generating visual concepts from structured inputs.
- Develop logic to automatically apply generated designs to a pre-defined layout template (e.g., packaging dieline).
- Ensure outputs meet basic specifications (aspect ratio, resolution, export format).
- Collaborate with product and design teams to fine-tune the prompt engineering and output handling process.
- Implement basic session management for user uploads and outputs.

Required Skills & Experience
- Proven experience with Python and frameworks like Streamlit, Gradio, or similar.
Hands-on experience with OpenAI tools, especially DALL·E 3 and ChatGPT APIs.
Experience with image manipulation libraries (e.g., PIL, OpenCV, or similar).
Familiarity with design principles, especially for print (e.g., CMYK, resolution, bleed).
Strong understanding of prompt engineering for creative outputs.
Ability to convert AI-generated images into structured templates or overlays (e.g., fitting onto dielines).
Comfortable working independently and iterating quickly based on feedback.

Nice to Have
- Knowledge of packaging design or print workflows
- Experience integrating other generative tools (e.g., MidJourney, Ideogram).
- Basic understanding of vector graphics (e.g., SVG generation).
- Exposure to UI/UX workflows for non-technical users.

What You’ll Be Building
- A functional MVP that enables users to generate branded design concepts from structured prompts and export those designs into formats suitable for further refinement or production use. Your work will serve as the AI foundation of a larger creative toolchain.

Skills: Python, Website Design, Graphic Design, User Interface / IA, HTML, ChatGPT, Streamlit, AI Image-to-text, AI Design, AI Development

Hello,

I’m looking for an experienced developer to create a bot for GTA RAGE MP that automates the mining job on a public server.

The bot should:
✅ Automatically detect and navigate to random mining job checkpoints (red markers visible both on the screen and minimap)
✅ Press the action key (E) to start mining when entering a checkpoint
✅ Detect when the checkpoint disappears and a mining progress UI appears
✅ Wait until mining is complete, then move to the next random checkpoint
✅ Periodically check inventory weight (by opening the inventory with key I and reading "XX/XX kg")
✅ When inventory is full:
- Navigate back to the van (marked as a car icon on the minimap)
- Unlock the van with F3, open the trunk using K → select van → trunk, deposit mined resources, then close trunk and lock van again
✅ Repeat the process indefinitely

Additional requirements:
- The bot must simulate human-like behavior (random delays, imperfect movements) to avoid detection
- No code injection or memory manipulation – should work externally via keyboard/mouse simulation and screen capture (PyAutoGUI, OpenCV or similar)
- Compatible with windowed fullscreen mode at 1920x1080 resolution
- Should run as a standalone application (.EXE preferred)

Optional:
- Auto-stop when receiving messages in chat or after a set number of hours
- Easy configuration for different keybinds or resolutions

Deliverables:
- Fully functional bot (source code and compiled .EXE)
- Instructions for setup and usage

Please apply only if you have prior experience developing bots for GTA (RAGE MP / FiveM) or similar games. Share examples of your previous work if possible.

Budget: Open to offers depending on your experience and delivery time.

Thanks!

Skills: Java, Python, C# Programming, Software Architecture, C++ Programming, Game Development, Programming, OpenCV, Automation, Bot Development

We are looking for expertise in Stereo Vision, IPC, Structure from Motion, Scene Graphs etc. - using both classical and deep learning based approaches. C++ and Python programming skills are required. Familiarity with deep learning using PyTorch is important. The work will require exploring existing frameworks and understanding recent developments. Good written english skills are necessary.

The goal is to develop software modules that use commonly available hardware (Livox/Ouster/Realsense etc.) and a combination of edge and cloud based compute (e.g. Jetson + AWS) to produce 3D models in real-time at a minimal computational cost.

This is a remote positionRequirements:

**- Bachelor's degree in Computer Science, Electrical Engineering, or related field; Master's degree preferred.- Proven experience working with computer vision libraries such as OpenCV, TensorFlow, or PyTorch.- Strong programming skills in languages like Python, C++, or MATLAB.- Familiarity with 3D reconstruction techniques like structure-from-motion (SfM), stereo matching, point cloud processing.- Experience with deep learning frameworks for image analysis tasks is a plus.**

Qualifications:

**- Excellent problem-solving abilities and analytical thinking skills.- Ability to work independently as well as part of a team in a fast-paced environment.- Strong communication skills to collaborate effectively with colleagues from diverse backgrounds.

I'm seeking an experienced AI or computer vision developer to help in building a web-based tool capable of measuring and extracting precise measurement data from real-time images or video.

The ideal candidate will have a strong background in image processing and machine learning, as well as experience in developing applications that handle and analyze live video or webcam feeds.

Your work will directly support the creation of a high-accuracy measurement system, similar in concept to pupillary distance scanners used by eyewear apps like Warby Parker.

Ideal skills:
Facial landmark detection (OpenCV, MediaPipe, Dlib, etc.)
Depth estimation or 3D facial modeling
Experience with precision measurement from 2D/3D input
Bonus: background in beauty tech, biometrics, or mobile camera scanning tools

• Please list any certifications related to this project
• Describe your recent experience with similar projects

We are seeking a skilled CV Developer to convert 2D images into 3D models using Python and OpenCV. The ideal candidate will have a strong background in computer vision and experience with 3D modeling. You will work closely with our team to ensure high-quality outputs and timely delivery. This project is expected to last for four months, with a budget in the range of $15,000 to $25,000. If you are passionate about computer vision and ready to tackle challenging tasks, we would love to hear from you!

• Describe a complex computer vision project you've completed. What made it technically challenging?
• What's your experience with document processing and extracting structured data from 2D images?
• Have you worked on projects involving 3D coordinate generation or mesh creation? Describe your approach.
• How do you typically approach a computer vision problem where you need high accuracy on varied input data?
• Rate your experience level (1-10) with: OpenCV, NumPy, 3D file formats, and Python optimization techniques.

I need computer vision models trained for image classification and detection using real-world datasets..

Key Requirements:
- Train model on specified datasets.
- Classify and detect various objects/features.
- Ensure high accuracy and reliability.
- Flask API development for trained models.
- Explainable AI (XAI)

Ideal Skills & Experience:
- Expertise in computer vision, especially with training models.
- Strong background in handling and processing real-world and microscopic images.
- Proficiency in relevant programming languages and frameworks (e.g., Python, TensorFlow, PyTorch).
- Experience in model evaluation and optimization.

Please provide a portfolio showcasing relevant projects and experience.

We’re looking for a computer vision developer to help build an AI-powered surveillance system that can detect firearms, track individuals in real time, and trigger automated responses through DJI drones.

The goal is to process video feeds from CCTV or drones, identify when a firearm is present, follow the subject, and send alerts with tracking data to external systems or law enforcement dispatch centers. This system will eventually connect to DJI Matrice drones using the Dock system.

Scope of work includes:
- Integrating or fine-tuning a pre-trained firearm detection model (YOLOv8, Roboflow, or similar)
- Connecting to RTSP or RTMP camera feeds
- Overlaying bounding boxes and detection confidence on live video
- Adding real-time tracking of individuals once a firearm is detected
- Outputting position data (pixel or GPS) to be used by a drone or mapping system
- Triggering DJI drone responses via SDK (FlightHub 2 or OSDK)
- Sending detection alerts via webhook or REST API
- (Optional) Building a lightweight dashboard that shows alerts and map-based tracking

Ideal candidate should have:
- Strong experience with Python, OpenCV, and PyTorch or TensorFlow
- Familiarity with YOLOv5 or YOLOv8, Ultralytics, or Roboflow pipelines
- Experience working with RTSP/RTMP streams
- Experience with object tracking (DeepSORT, ByteTrack, or similar)
- Knowledge of DJI SDKs and drone integration is a big plus
- Background in surveillance, security, or similar AI applications is preferred

We’re aiming to complete the first build in next few weeks. There is potential for long-term work as the system evolves.

If this sounds like something you’ve built before, please apply with a few examples of related work.

I'm looking for a computer vision expert to optimize a video processing script. The primary goal is to increase speed.

Key focus areas include:
- Optimise video processing (right now it takes 3 minutes to process 1 second video, should get it down to almost 30 second)
- Algorithm efficiency

The current code is written in Python.

Ideal skills and experience:
- Proficiency in Python
- Strong background in computer vision
- Experience with video processing and optimization techniques
- Familiarity with handling frame rates and improving algorithm efficiency
- Expert in multi-processing / multi-threading
- Expert in openCV
- Expert in AI models like Roboflow, Yolo, Kmeans and other models.

Please provide examples of similar work done.

Skills: Python, Matlab and Mathematica, Software Architecture, Machine Learning (ML), Image Processing, OpenCV, Video Processing, Computer Vision, Deep Learning, YOLO