Mayukh Deb

torch-dreams. Flexible Feature visualization on PyTorch, for research and art :mag_right: :computer: :brain: :art:

differentiable-morphogenesis. experimenting with differentiable models of morphogenesis :microscope: :microbe:

rover. Reverse engineer your pytorch vision models, in style

mind_the_bend. vision based deep learning for racing games :racing_car:

tiny-transformer. tiny transformer(s) on pytorch - I made this while trying to understand how they work

deep-chicken-terminator. tracking wildlife in minecraft using deep learning :hocho: :chicken:

deep-chicken-saviour. using adversarial attacks to confuse deep-chicken-terminator :shield: :chicken:

torch-dreams-notebooks. torch-dreams exploration notebooks

ECG-heartbeat-categorization. deep learning to obtain medical inferences from ECG data :chart_with_upwards_trend:

adventures-with-augmentation. Exploring and experimenting with microscope imagery datasets :microscope:

vaporizing-villagers. Killing minecraft villagers with the help of image segmentation

RAFT-opticalflow-runwayML. Recurrent All Pairs Field Transforms for Optical Flow: on RunwayML

math-stuff. me teaching some mathematical concepts to myself

gradient-games. mostly contains scratchy numpy code to build and train basic neural networks. You might also find a few experiments on torch.

exploring-GANs. Experiments with GANs, and some stuff I learned on the way

worm-embryo-classifier. Flask webapp to determine the embryogenic stage of the C.elegans worm using deep learning :microscope:

simple_face_ID. collects cropped facial data from webcam -> trains CNN -> classifies faces from live webcam feed

loss_landscape. Visualizing the lower dimensional error space of various datasets on deep neural networks :mountain:

deepfakes. Figuring out how to generate deepfakes and making notes on the way

auroraGAN. northern lights but with GANs

shoeGAN. Generating interpolations between random shoes with GANs

torch-dreams-runwayML. model server template for torch-dreams on runwayML

tgtqdm. who needs work-life balance when you can watch your logs on the phone?

pascal-voc-caffemodel-pytorch. Port of the original Pascal VOC 2012 multilabel classification model from caffe to pytorch

Mayukhdeb.

traffic_eye. nn to detect and interpret traffic signs :vertical_traffic_light:

Mayukhdeb.github.io. CSS

reverse-engineering-segmentation-models. turning a face segmentation model inside out

sneaker_vision. scrapes google images -> normalises them -> feeds into NN -> NN trains on the data

exploring-NLP. bunch of notebooks where I explain NLP to myself

patrick. Tiny neural net library written from scratch with cupy :warning: under construction :warning: