Browse categories
Explore
Fiverr Pro
English
$
USD
I will build custom computer vision models for object detection and image analysis
About this Gig
Need a computer vision model that works on your real images, not just a demo dataset? Khan's AI is an applied AI studio registered with Join IT KPK, Pakistan. We build, train and deploy vision models for businesses.
What we can build for you:
- Object detection and counting (YOLO): products, vehicles, people, defects
- Image classification and quality inspection for manufacturing
- Segmentation for medical, satellite or industrial images
- OCR and document scanning: invoices, forms, ID layouts, drawings
- Face, pose and gaze tracking with MediaPipe and OpenCV
- Edge deployment: Raspberry Pi, Jetson and mobile
How we work:
- You share sample images and the goal.
- We confirm feasibility and the metric we will be judged on.
- We label, train and test on data the model has never seen.
- You get the model, clean Python code, an inference script and a short report.
Why clients choose us:
- Honest evaluation with no inflated accuracy numbers
- You own the code and the model
- NDA available for private data
- Clear updates at every milestone
Not sure which package fits? Send us a message with a few sample images and we will reply with a plan.
Programming language:
Python
•
R
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
•
Excel
•
Colab
•
PyTorch
Frameworks:
Scikit-learn
•
Keras
•
PyTorch
•
Panda
My Portfolio
Other Data Science & ML Services I Offer
FAQ
How many images do I need?
It depends on the task. Many detection projects start well with 200 to 500 labeled images per class. If you have fewer, we can use transfer learning and augmentation, and we will tell you honestly what accuracy to expect.
What do I receive at the end?
The trained model weights, Python source code, an inference script, setup instructions and a short results report with test metrics.
Do you sign NDAs?
Yes. We are happy to sign your NDA before you share any private data.
Can you label my data?
Yes. Labeling can be added as an extra, or we can guide your team to label with our annotation template.
Can the model run on a phone or small device?
Yes. We can export to ONNX, TFLite or TensorRT and optimise it for Raspberry Pi, Jetson or mobile apps.
