I will do object detection, object tracking and computer vision with yolo

I train object detection models that work on your actual footage, not just clean benchmark images.

Most detection projects fail for one reason: the model was validated on tidy data and then deployed on real-world video with motion blur, bad lighting, occlusion, and unusual angles. I build for that from the start.

What I do:

• Custom object detection with YOLO, trained on your dataset
• Object tracking across video frames with consistent IDs
• Image classification and segmentation
• Counting, zone detection, proximity and risk logic
• Dataset annotation and augmentation where you have no labels yet
• ONNX export and deployment support

Why me:

• 5+ years building production computer vision, not tutorials
• Assistant Manager Research at Pakistan's Centre for Language Engineering
• Visiting researcher, University of Konstanz
• LRECpaper published; public dataset and trained detection model released on Hugging Face
• 20 five-star reviews with repeat clients

Recent work: I built and released a complete traffic detection stack for mixed-traffic environments: 13 hours of source footage, 1,506 annotated boxes, a fine-tuned YOLO model at 0.72 mAP50, plus a proximity risk overlay app. Dataset and model are bot