Browse categories
English
$
USD
Connect to your AI
I will build deep learning image classification models with tensorflow
AI Engineer and Data Scientist
I'm Seher, an AI engineer and data scientist at Datix AI, with a BSc in Data Science from the University of Kotli.
I help businesses and researchers turn raw data into clear answers, through statisti...
About this Gig
I build deep learning models that sort images into the categories you need, such as product types, defects, crops or land use.
What I do:
Prepare and augment your images, use transfer learning with proven CNN backbones (ResNet, EfficientNet) in TensorFlow and Keras, tune the model, and test it on images it has never seen.
What you get:
The code, the trained model, accuracy results with a confusion matrix, and clear instructions for running it on new images.
Why work with me:
I am an ML Engineer and data scientist at Datix AI. My recent work includes a satellite land-use classifier on 27,000 Sentinel-2 images (93% test accuracy) and a 10-class product image classifier on 70,000 images (90% accuracy).
Send me a sample of your images, and I will tell you honestly whether you have enough data before we start.
Expertise:
Image processing
•
Feature learning
•
Classification
Programming language:
Python
•
Colab
Tools:
Jupyter Notebook
•
OpenCV
•
TensorFlow
Frameworks:
Scikit-learn
•
Keras
My Portfolio
FAQ
How many images do I need?
With transfer learning, around 100 to 300 clear images per class is often enough to start. Send me a sample and I will tell you honestly before we begin.
Can the model run on my own computer or server?
Yes. I deliver the saved model and a simple Python script to run it on new images. Premium includes a lighter export for faster use on normal hardware.
What accuracy can you promise?
Accuracy depends on your images and classes, so I never promise a number upfront. I report honest test accuracy on unseen images and explain where the model makes mistakes.
