My projects

Image Completion using Deep Convolutional Generative Adversarial Network

Adversarial training is a way to train two neural networks simultaneously. The first one is the Discriminator that takes an input (e.g. an image) and outputs a scalar indicating whether the image looks “natural” or not. The second network is a Generator. The role of this generator is to generate a fake image so as to train the Discriminator to output the correct probability of it being “fake” or not. During the process, the Generator network learns to generate natural looking images (face images in this case) and the Discriminator learns to distinguish between the natural and fake images.

I’m applying this DCGAN architecture in image completion/inpainting. Some preliminary results are available in the project URL.

Implemented in Python using TensorFlow.

URL: https://github.com/saikatbsk/ImageCompletion-DCGAN

TensorFlow-GAN: Generative Adversarial Network

tf-gan

This is a basic GAN implementation with an aim to understand the core concepts behind "generative models via an adversarial process". After training on the MNIST dataset, the generator comes up with images that resemble handwritten digits.

Implemented in Python using TensorFlow.

URL: https://github.com/saikatbsk/TensorFlow-GAN

DeepView

DeepView is an approach for classifying images using Convolutional Neural Networks. Here, I have explored model scalability, optimization techniques, activation functions and other important aspects of deep neural network architectures. The final model has currently been tested on the CIFAR-10 dataset with promising results. I plan to explore data augmentation and see how that affects the classification accuracy.

Implemented in Python using Keras.

URL: https://github.com/saikatbsk/DeepView

MNIST-CNN

mnist

CNN model trained to classify handwritten digits from the MNIST dataset. MNIST is a dataset of 60,000 28x28 grayscale images of the 10 digits, along with a test set of 10,000 images. The model achieves 99.23% accuracy which is at par with some of the current benchmarks.

Implemented in Python using Keras. Data visualization using t-SNE.

URL: https://github.com/saikatbsk/MNIST-CNN

Understanding-CNN

understanding-cnn

Visualizing ConvNet layers provide a deep Understanding of how neural networks see the world. Maximizing the responses from conv filters give us a neat visualization of the convnet's modular-hierarchical decomposition of its visual space. The first layers basically just encode direction and color. These direction and color filters then get combined into basic grid and spot textures. These textures gradually get combined into increasingly complex patterns. The filters become more intricate as they start incorporating information from an increasingly larger spatial extent.

Implemented in Python using Keras, Scipy.

URL: https://github.com/saikatbsk/Understanding-CNN

Vincent: AI Artist

ai-artist

Style transfer is the technique of recomposing images in the style of other images. Vincent is an implementation of the popular neural-style paper.

Implemented in Python using Keras, Scipy, and Sklearn.

URL: https://github.com/saikatbsk/Vincent-AI-Artist

DeepDream

deep-dream

Baking my own Deep Dream. This is an implementation of the popular conv filter visualization technique known as DeepDream/Inceptionism introduced in Google Research Blog.

Implemented in Python using Keras, Scipy.

URL: https://github.com/saikatbsk/DeepDream

bagOfFeatures

bof

Image classification using the Bag-of-Features model. Training images are represented using clustered and quantized Speeded-Up Robust Features (SURF). Classification using K-Nearest Neighbor.

Implemented in Octave.

URL: https://github.com/saikatbsk/bagOfFeatures

Wave: A Perceptive User Interface

A perceptive user interface based on real-time hand detection and tracking. Requires a webcam. Useful for applications such as gesturebased point and click interfaces.

Implemented in C++ using OpenCV.

URL: https://github.com/saikatbsk/Wave

ISHARA: Mouse Control with Gesture

An interface for controlling mouse pointer using finger gestures. Requires a webcam. Includes gestures for emulating click (left and right) and scroll using color markers.

Implemented in C++ using OpenCV, and Qt.

URL: https://github.com/saikatbsk/Ishara

I’ve also worked on some minor and fun projects,

GameOfLife

game_of_life

Simulating Conway's 'cellular automaton'. Implemented in Octave.

URL: https://github.com/saikatbsk/GameOfLife

tinyOS

A tiny operating system. Written in C and Assembly.

URL: https://github.com/saikatbsk/tinyOS

CrappyBird

crappybird

Yet another flappy bird clone. Written in JavaScript using p5js.

URL: https://github.com/saikatbsk/CrappyBird

RacingMania

racingmania

Retro car racing game. Written in JavaScript using p5js.

URL: https://github.com/saikatbsk/RacingMania

rss facebook twitter github youtube mail spotify instagram linkedin google google-plus pinterest medium vimeo stackoverflow reddit quora