Browse categories
Explore
Fiverr Pro
English
$
USD
I will build custom yolo object detection and computer vision models
About this Gig
Need a model that can spot something specific in your images or video? I build custom object detection systems that run in production, not just in a notebook.
WHAT YOU GET
- A detection model trained on your object classes
- Inference script and full source code
- An accuracy report (precision, recall, mAP) so you know what you're getting
- Higher tiers add data preprocessing, docs, cloud deployment and a REST API
ARCHITECTURES
YOLO (v5-v11), Faster R-CNN, SSD MobileNet, ResNet, EfficientNet
TOOLS
PyTorch, OpenCV, Ultralytics, Keras, ONNX, ByteTrack, DeepSORT, FastAPI, Docker, AWS
LANGUAGES
Python, SQL
WHAT I NEED FROM YOU
- Images or video of the objects you want detected
- For Basic: annotations in YOLO or COCO format
- A list of the classes to detect
- Any accuracy or speed target you have
Unlabelled data is fine - add the Image annotation extra. Basic does not include cloud deployment, API integration or hosting.
Message me before ordering with a few sample images and what you need detected. I'll tell you honestly whether it's doable and which package fits.
FAQ
Do I need to provide labelled data?
For Basic, yes - annotations in YOLO or COCO format. If your images aren't labelled yet, add the Image annotation extra (covers up to 300 images) and I'll handle it. For larger sets, message me and I'll send a custom offer.
How many images do I need?
As a rough guide, 100-200 labelled images per class gives a usable model and 500+ gives a strong one. Fewer can work if the objects are visually distinct. Send samples and I'll tell you honestly.
What accuracy will I get?
It depends on your data - image quality, class balance, how similar the classes look. I won't promise a number before seeing samples. What I will do is send a full precision, recall and mAP report so you know exactly what you have.
Can you deploy the model for me?
Yes. Premium includes cloud deployment and a REST API, or you can add either as an extra to Basic or Standard.
Will it run in real time?
Usually yes - YOLO runs comfortably above 30 FPS on a GPU. If you need it on CPU or an edge device, add the Speed optimization extra and tell me your target hardware.
What if the accuracy isn't good enough?
Every package includes revisions. If the model underperforms on the data you supplied, I'll retrain and tune within your revision count. If the dataset itself is the problem, I'll tell you exactly what needs to change.
