Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a custom computer vision model for detection, tracking, or segmentation
Athar
About this gig
Are you building a drone monitoring system, security camera app, or any product that needs to detect, track, or count objects in images or video? I can build that for you.
I'm a computer vision engineer hands-on experience building real systems including a drone traffic analyzer that counts vehicles across user-defined lines using YOLOv8 + ByteTrack with custom ReID logic, and a flood area segmentation models.
What I can build for you:
Custom object detection (YOLOv8, fine-tuned on your data)
Multi-class image segmentation
Object tracking in video (ByteTrack, DeepSORT)
Vehicle, person, or custom class counting
Web interface to upload images/video and see results live
Inference scripts ready to deploy
All deliverables include clean Python code, documentation, and a results report.
I've worked with: YOLOv8, OpenCV, PyTorch, FastAPI, MediaPipe, LSTM, TensorFlow.
Message me before ordering if you're unsure which package fits your needs or I will make a custom order just for you.
Cheers...
Get to know Athar
Athar
I will build a custom computer vision model for object detection, tracking, or s
- FromBangladesh
- Member sinceMay 2026
- Avg. response time1 hour
Languages
Bengali, English, Urdu, Hindi
CV & full-stack engineer, from Bangladesh. I build real systems — not just notebooks.
Projects: Traffic analyzer (YOLOv8 + ByteTrack + ReID), flood segmentation model, HAR with LSTM + MediaPipe, and run a popular website protarok.org — a scam-archiving platform covered by 4 national news outlets.
Clean, documented, production-ready code. Model training, full pipelines, or a complete web app around your CV system — I handle all of it.
Stack: YOLOv8 · PyTorch · OpenCV · FastAPI · React · Node.js · PostgreSQL
Message me before ordering — I'll tell you honestly if I can help.
My Portfolio
FAQ
Do I need to provide a dataset?
For custom training (Standard/Premium), yes — ideally 200+ labeled images per class. If you don't have one, I can advise on using public datasets or data augmentation to work with less. For pre-trained models (Basic), no dataset needed.
What format will the deliverables be in?
Python scripts, trained model weights (.pt), and a PDF/markdown report showing accuracy metrics (mAP, precision, recall). For Premium, a working web app you can run locally. Or i can deploy it for you for some extra payments.
Can you work with video, not just images?
Yes. I can process video files frame-by-frame and return an annotated output video with bounding boxes, class labels, and optional object counts.
What if I need something not listed in the packages?
Message me. Most custom requests (specific model architectures, edge deployment, real-time webcam, etc.) can be done as a custom offer.
