Modassir Afzal's Projects
Implementation from scratch of [Arbitrary Style Transfer in Real-time with Adaptive Instance Normalization [Huang+, ICCV2017]]
It is a repository that is a collection of algorithms and data structures with implementation in various languages.
The open source version of the Amazon Rekognition docs. You can submit feedback & requests for changes by submitting issues in this repo or by making proposed changes & submitting a pull request.
Anime Face Generation using GANS and Label Smoothing.
ResNet-34 Model trained from scratch to classify 450 different species of birds with 98.6% accuracy.
Image classification Neural network made from scratch
I have trained two different CNN models for binary image classification to see which architecture has better accuracy, takes less time in training, how hyperparamters affect training and how many epochs do each of them need. I achieved 96% accuracy on the best model.
Implementation of [Image-to-Image Translation with Conditional Adversarial Networks [Isola+, CVPR2018]]
A SegNet model trained for segmentation of Lanes suitable for driving for automobiles.
Badges for your personal developer branding, profile, and projects.
Config files for my GitHub profile.
A CNN model trained on 50,000 images for classification of images on 10 different classes.
Probability Distribution Functions is a Python package to help plot, calculate distributions of data using different probability distribution types and formulas.
All Algorithms implemented in Python
I'll be putting up simple python tools I made. Feel free to suggest changes to make it better :)
Implemented the Deep Residual Learning for Image Recognition Paper and achieved better accuracy by customizing different parts of the architecture.
Implemented Deep Residual Learning for Image Recognition Paper and achieved lower error rate by customizing different parts of the architecture.
Designed a smaller architecture implemented from the paper Deep Residual Learning for Image Recognition and achieved 93.65% accuracy.
The top 5% of the titanic competition in Kaggle. achieved this through ensemble of models