Parikshit Kumar's Projects
A web-app that uses a vanilla generative adversarial network and unsupervised learning to generate random 28X28x1 plasma style images of digits from 0 to 9
A basic OCR that uses Tesseract and OpenCV
Has you play a best of one Rock Paper Scissors game vs the Computer using OpenCV and Convolutional Neural Networks to detect rock, paper and scissors move.
Binary classifier CNN model from scratch using PyTorch, classifies Bananas and Pineapples
Finds 3 largest contours in a given image
Uses a resnet to classify animal images into 10 classes
An OpenCV application that uses YOLOv3 object detection and Darknet weights to detect firearms in a given image, or video.
Generative Modeling using GANs, DCGANs, WGAN, WGAN-GP, ProGAN, Conditional GANs and Controllable Generation.
Explaining Git and GitHub.
Uses the computer webcam to interpret hand gestures for various tasks like sign language recognition and automation using CNNs.
A web-app based on Wasserstein Generative Adversarial Network architecture with GP that generates multiple realistic paintings, trained on 8k Albrecht Dürer's paintings, includes super-res mode.
Using Pix2Pix GAN for translating Anime images to something more aesthetic
Classifies cell images into Parasitized/Uninfected using a CNN
starterdex, but better, an app that identifies Pokémon by using densenet201 and fastai over PyTorch, knows about all 151 generation one Pokémon + 15 mega evolutions and tells you all about them.
An OpenCV application that uses YOLOv3 and YOLOv3-Tiny object detection and weights trained on a custom dataset to detect firearms in a given image, video and in real-time.
intro
Classifies X-Ray images into pneumonia/no pneumonia
Checking out PL and fastai here
PyTorch Tutorials from Jovian.ai for Deep Learning
easy OpenCV OCR implementation from scratch
Resnet architecture to classify between rock, paper and scissors class dataset from Kaggle
Some scripts and data collection tools
A computer vision web-app that uses deep learning and Residual Neural Networks to identify your Pokémon and then tells you all about it.
Prototype object detection project using yolov3-tiny trained on the coco dataset