Browse categories
Explore
Fiverr Pro
English
$
USD
I will build a custom ai image classification model in python
Jan Ali
About this gig
I build and train custom computer vision models that classify or detect things in images product photos, quality checks, medical/research imaging, document sorting, or any custom visual task.
WHAT I'VE BUILT:
Trained and deployed a skin lesion classifier (MobileNet-V3, ISIC dataset, 83% validation accuracy) full pipeline from training to a live web app. Details available on request.
WHAT I CAN BUILD:
- Custom image classification
- Basic object detection
- Fine-tuning pre-trained models on your data
- Honest accuracy reporting, no inflated claims
- Optional simple web app for non-technical use
WHY ME:
CV student, part of a small dev team (Nexoratec). I'll tell you upfront if your dataset is too small for reliable results before you pay, not after.
PROCESS:
1. Share your images + what you want classified
2. I confirm feasibility and scope first
3. I train, evaluate honestly, and deliver the model with a clear report
WHO THIS IS FOR:
Startups, researchers, or businesses with image data who want it sorted or checked automatically instead of manually.
Message me with a data sample before ordering Standard/Premium for an accurate scope.
Get to know Jan Ali
Jan Ali
Transforming Ideas into Stunning Visuals Your Vision My Design
- FromPakistan
- Member sinceAug 2022
- Avg. response time6 hours
Languages
English, Urdu
I am Jan Ali — a Computer Vision Engineer and AI student building real-world solutions that go far beyond tutorials. I specialize in Python-based CV systems including background removal, face detection, OCR pipelines, and deep learning image classifiers.
I am currently building a portfolio of 30 Computer Vision projects — each one solving a real problem with production-level code. My work is publicly available on GitHub and deployed live for real users.
I deliver clean code, clear communication, and results that actually work.
My Portfolio
FAQ
How many images do I need to provide?
Minimum ~50-100 images per category for a basic model; more gives better accuracy. I'll tell you honestly if your dataset is too small.
Will the model be 100% accurate?
No model is perfect I'll give you a real accuracy number from testing, not a sales promise. I'll also tell you ways to improve it.
Can you use medical/sensitive images?
Yes, but I don't provide medical diagnoses only the technical model. Any health-related use must include a clear disclaimer that it's not a substitute for professional diagnosis.
