You will get A Custom YOLO and Object Tracking Pipeline in Python
Project details
> Looking for a production-ready Computer Vision pipeline that doesn't lose track of objects in crowded spaces? This service delivers a high-accuracy YOLO detection and ByteTrack multi-object tracking (MOT) pipeline built natively in Python.
> Standard detectors lack memory—if an object is briefly blocked, they lose its identity, causing duplicate counts. This pipeline solves that by combining ultra-fast YOLO (v8/v11) inference with ByteTrack's robust association matrices to maintain persistent IDs, handle complex occlusions, and predict motion trajectories.
> Core Deliverables:
> Tracking & Re-ID: Unique, persistent IDs for people, vehicles, or custom objects.
> Line Crossing & Analytics: Virtual tripwires for entry/exit counts and custom polygon zones to track dwell times.
> Data Logging: Automation scripts exporting tracking coordinates directly to structured JSON/CSV logs.
> Why Choose Me?
> I bring real-world startup engineering to your project. I built the core tracking engine for a sports analytics startup, helping them secure initial funding and sign 3 MoUs with national- level clubs. I deliver clean, modular, and hardware-optimized Python code.
> Standard detectors lack memory—if an object is briefly blocked, they lose its identity, causing duplicate counts. This pipeline solves that by combining ultra-fast YOLO (v8/v11) inference with ByteTrack's robust association matrices to maintain persistent IDs, handle complex occlusions, and predict motion trajectories.
> Core Deliverables:
> Tracking & Re-ID: Unique, persistent IDs for people, vehicles, or custom objects.
> Line Crossing & Analytics: Virtual tripwires for entry/exit counts and custom polygon zones to track dwell times.
> Data Logging: Automation scripts exporting tracking coordinates directly to structured JSON/CSV logs.
> Why Choose Me?
> I bring real-world startup engineering to your project. I built the core tracking engine for a sports analytics startup, helping them secure initial funding and sign 3 MoUs with national- level clubs. I deliver clean, modular, and hardware-optimized Python code.
Machine Learning Tools
Keras, MLflow, NumPy, Open Neural Network Exchange, OpenCV, Python, PyTorch, SciPy, TensorFlowWhat's included
| Service Tiers |
Starter
$249
|
Standard
$649
|
Advanced
$1,399
|
|---|---|---|---|
| Delivery Time | 4 days | 7 days | 19 days |
Number of Revisions | 1 | 2 | 3 |
Number of Model Variations | 1 | 1 | 2 |
Number of Scenarios | 1 | 2 | 3 |
Number of Graphs/Charts | 0 | 1 | 1 |
Model Validation/Testing | - | ||
Model Documentation | - | ||
Data Source Connectivity | - | - | |
Source Code |
Optional add-ons
You can add these on the next page.
Additional Revision
+$75
Additional Model Variation
(+ 2 Days)
+$200
Additional Scenario
(+ 1 Day)
+$150
Additional Graph/Chart
(+ 1 Day)
+$100
Model Validation/Testing
(+ 1 Day)
+$150
Model Documentation
(+ 1 Day)
+$100
Data Source Connectivity
(+ 3 Days)
+$350Frequently asked questions
About Own
Computer Vision Engineer
Amman, Jordan - 12:56 am local time
I help companies build Computer Vision systems that move beyond proof-of-concept and deliver reliable performance on real video data.
SERVICES I DELIVER
COMPUTER VISION & TRACKING SYSTEMS
▪ Object detection and multi-object tracking
▪ Trajectory analysis and motion modeling
▪ Custom Computer Vision pipelines using Python and OpenCV
VIDEO INTELLIGENCE & AUTOMATED INSIGHTS
▪ analytics and event detection
▪ Automated video understanding
▪ Performance analysis and reporting systems
▪ Feature extraction and AI-powered decision support
PRODUCTION AI ENGINEERING
▪ Machine Learning and Deep learning model development and optimization
▪ FastAPI-based AI services
▪ Dockerized deployments
▪ End-to-end ML pipeline development
TECHNICAL EXPERTISE
Computer Vision & Tracking:
YOLO, ByteTrack, DeepSORT, OpenCV, Multi-Object Tracking, Kalman Filtering & Trajectory Smoothing, Video Analytics & Event Detection
Machine Learning & AI:
PyTorch, TensorFlow, Scikit-Learn, Deep Learning, Feature Engineering
Backend & Deployment:
FastAPI, Docker, REST APIs, Google Cloud
Performance & Data Processing:
Python, Python Multiprocessing, SQL, Data Analysis & Visualization
RELEVANT PROJECT IMPACT: SOCCER VIDEO ANALYTICS PLATFORM
Led the development of a soccer video analytics AI system, designing and building the complete Computer Vision architecture before collaborating with the backend team to integrate and deploy the system.
Key engineering outcomes:
▪ Designed an end-to-end analysis pipeline using YOLO for detection and ByteTrack with Kalman-based smoothing for tracking, significantly improving identity consistency and reducing tracking drift across challenging match footage.
▪ Developed custom physics-based heuristics using trajectory analysis, speed, acceleration, angle variation, and motion patterns to automate pass detection, ball airtime detection, and gameplay event extraction.
▪ Built advanced tactical visualization tools including Voronoi diagrams, trajectory overlays, and radar views, enabling coaches and analysts to interpret player positioning, movement patterns, and team shape more effectively.
▪ Implemented coordinate transformation and pitch mapping techniques to convert raw video detections into actionable spatial analytics and tactical insights.
▪ Optimized the video processing pipeline using Python multiprocessing, reducing processing latency by more than 50% on an RTX 3060 while maintaining analysis quality.
▪ Designed the system for production deployment as a FastAPI microservice, enabling integration with backend applications and scalable AI inference workflows.
▪ Combined Computer Vision, tracking, event detection, and analytics into a unified platform capable of transforming raw match footage into structured performance data and automated post-match insights.
WHY YOU SHOULD HIRE ME
Many developers can train a model. Fewer can build the infrastructure, tracking logic, data pipelines, and engineering systems required to make that model useful in production.
My background combines Computer Vision, Deep Learning, Data Science, and software engineering, allowing me to solve problems across the entire AI lifecycle—from raw video ingestion to automated insights and deployment.
WHAT YOU CAN EXPECT
✓ Solutions designed for real-world video conditions, not just benchmark datasets
✓ Strong focus on tracking reliability, performance, and scalability
✓ Clear communication and collaborative development process
✓ Clean, maintainable, production-ready code
✓ End-to-end ownership from model development to deployment
If you're building a Computer Vision product and need more than just a trained model—whether it's object tracking, video intelligence, event detection, analytics, or production deployment—I'd be happy to discuss your project.
Happy to answer any questions you may have.
Steps for completing your project
After purchasing the project, send requirements so Own can start the project.
Delivery time starts when Own receives requirements from you.
Own works on your project following the steps below.
Revisions may occur after the delivery date.
Source Video & Constraint Review
I analyze your sample video files or stream feeds to evaluate the camera angle, lighting, and expected occlusion density. We finalize the coordinate boundaries for line-crossing or zone tracking before any code is written.
Pipeline & Tracker Assembly
I build the core Python script integrating OpenCV and the YOLO model with the multi-object tracking engine. Every target object is assigned a unique, persistent ID, and the system is calibrated to handle overlapping objects smoothly.