Name: Ratnajit Mukherjee
Type: User
Company: INESCTEC, University of Porto
Bio: I am Computer Vision scientist working as a research fellow at INESCTEC Portugal and an HDR image/video processing and deep learning enthusiast.
Location: Vila Real, Portugal
Blog: http://warwick.academia.edu/RatnajitMukherjee
Ratnajit Mukherjee's Projects
A demo project showing the effect of Gamma Curves on Images
Lightweight models for real-time semantic segmentationon PyTorch (include SQNet, LinkNet, SegNet, UNet, ENet, ERFNet, EDANet, ESPNet, ESPNetv2, LEDNet, ESNet, FSSNet, CGNet, DABNet, Fast-SCNN, ContextNet, FPENet, etc.)
Just a basic demo to familiarize
Emotion classification has always been a very challenging task in Computer Vision. Using the SSD object detection algorithm to extract the face in an image and using the FER 2013 released by Kaggle, this project couples a deep learning based face detector and an emotion classification DNN to classify the six/seven basic human emotions.
Solution to extract (debayer) multiple exposures from Flare (IO Industries) cameras and merge them into HDRs.
HDR image reconstruction from a single exposure using deep CNNs
A Keras implementation of CenterNet with pre-trained model (unofficial)
Pytorch-SSD
A Keras port of Single Shot MultiBox Detector
The Tiny ImageNet project started as a complementary real-life classification problem with Stanford Computer Vision Course CS231n whereby the project consists of a subset of the ImageNet dataset and a competitive chart is maintained.
Uniform Color Space based HDR video compression: This repository contains the code for a novel HDR video compression algorithm which has been proposed to compress HDR video frames to codec suitable YUV files which can be compressed using 10-bit video codecs (x264/x265/AV1)