Door categorieën bladeren
Ontdekken
Fiverr Pro
Nederlands
$
USD
I will train a custom yolo object detection model on your dataset
Computer Vision Engineer, Sports and Retail Video Analytics
Hi, I'm Raza. I've been building computer vision systems for 7+ years, and these days I work with a intelligent team of CV engineers on sports and retail video.
I've tracked tennis rallies, built an...
Gescreend door Fiverr Pro
Muhammad Raza is geselecteerd door het team van Fiverr Pro vanwege diens expertise.
Gescreend voor
Data science en ML
Over deze dienst
Generic object detection models look great in demos and then miss half the things you care about on your own images. The fix is almost always the same: train on your data, for your objects, in your lighting.
That's what I do. You send images or video, I clean and label the data if needed, train a YOLO model, and show you exactly how well it performs per class with real test results, not just a single accuracy number.
A few things I've built:
- PCB defect detection that marks scratches and missing components
- Box counting on a conveyor, zone by zone
- Rice grain quality inspection
- PPE and safety detection across multiple warehouse cameras
You get model weights ready to use, exported to ONNX, TFLite, CoreML or TensorRT depending on where it will run: a server, a PC, an NVIDIA Jetson or a phone.
Send me a few sample images and tell me what you want detected before ordering. I'll tell you honestly if your data is enough or if we need more.
Programmeertaal:
Python
•
Colab
•
Amazon SageMaker
•
Overige
Frameworks:
Scikit-learn
•
Google ML Kit
•
keras
•
PyTorch
•
Overige
Mijn portfolio
Veelgestelde vragen
What do you need from me?
Some sample images or a short video, the list of objects you want detected, and where the model will run. If your images are already labelled, even better.
Do I need thousands of images?
Not always. It depends on how different your objects look and how much they vary. For simple cases a few hundred images per class can work. I check your data first and tell you honestly.
Can you label my images?
Yes. Labelling is included in Standard and Premium. For Basic, I work with images you have already labelled.
Where can the model run?
Anywhere you need. I export to ONNX, TFLite, CoreML or TensorRT, so it can run on a server, PC, NVIDIA Jetson, Android or iPhone.
Can it also track and count objects?
Yes. In Premium I add object tracking so each item keeps an ID across frames, which is useful for counting and movement analysis.
