Arpan Gupta's repositories
KTH_Optical_Flow
This repository contains code supporting the paper "Action Recognition from Optical Flow Visualizations" by Arpan Gupta and M. Sakthi Balan.
CricketShotDetection
PhD Thesis project on Cricket shot detection.
batsman_detection
Detecting Batsman by finetuning Faster RCNN in PyTorch
batsman_pose_track
Track a batsman using the person poses given for the sequence of frames.
BOW_action_recognition
Project to recognize actions in videos based on Bag of Words model.
cluster_strokes
Cluster the Cricket strokes to determing the different stroke categories in an Unsupervised Learning Framework.
CricketStrokeLocalizationBOVW
Using BOVW with GRU / LSTM RNN networks for temporal Cricket stroke localization from untrimmed videos
KTH_action_recognition
Action Recognition for KTH dataset using person motion features.
localization_finetuneC3D
Finetune a C3D model on the highlights videos and then on the main dataset training samples.
StrokeAttention
Applying Soft Attention to Cricket Strokes in telecast videos
CricketStrokeLocalization
A repository for holding files for Temporal Cricket Stroke Localization on Untrimmed Videos (advanced).
InstallerScripts
This repository contains scripts for installation of open source tools like hadoop, OpenCV, Spark etc. The scripts are not supposed to be executed 'as-is' but to be taken as a reference only and may be customized as per your system settings.
TrackNetv3
TrackNet for Cricket Ball Tracking
Keccak_CortexM4F
This repo is part of my M.Tech. thesis which implements SHA-3 hash function (Keccak) on CortexM4F microcontroller. The code for the desktop interface is in Keccak_Interface repo.
Keccak_Interface
This repo is the second part of my M.Tech. thesis work. The code for first part is in Keccak_CortexM4F repo.
localization_gru
This repository contains the code for "Dataset Creation and Benchmarking for Cricket Stroke Localization from Untrimmed Videos"
localization_rnn
Cricket Action Localization using sequential deep neural networks and features extracted using pretrained models.
StrokeSelfSupervision
Learning Cricket Stroke Representations using Self-supervised approaches.
TopicModel
Applying topic models for unsupervised activity recognition in videos.
TRN.pytorch
A PyTorch reimplementation for the ICCV 2019 paper "Temporal Recurrent Networks for Online Action Detection" applied to Cricket Stroke Localization.