Journals and Patents
Project 01
Human Identification Using Gait analysis and 3D Convolutional Network under supervision of Dr.P.Supraja.
This Project is about human recognition by their gait i.e. walking style by analyzing their gait features. It includes Object detection, Silhouette extraction, Skeletonisation, 3D CNN, recognition.
It took one day to preprocess the 18000 videos of 181 objects from 11 different angles and 8 days to train the network to recognize the object i.e. human by his/her gait features.
Research Paper: Applied.
Patent: Published in IPindia 01/2020,1, Pg- 208
Project 02
This is the caption generated by Encoder and Decoder module of Video Captioning
Video frames as input given to the model
Graph obtain on training video captioning model
This is the caption generated by Encoder and Decoder module of Video Captioning
Video Captioning Using Encoder and Decoder Module under supervision of Dr.P.Supraja
In this project, we used 3D CNN with LSTM to train the neural network to generate the caption on the basis of the features present in the frames. We word embedding and vectorization for sanitization of the captions on which we trained our neural network.
After training for five continuous days we got the captions which were 92% relevant to the frames of the video.
Patent: Published in IPindia 51/2019,1, Pg- 61314
Project 03
Silhoutte of human waliking
Trainning Epochs of Model
Graph of trained model
Silhoutte of human waliking
Enhanced Human Gait Prediction under supervision of Dr.P.Supraja
In this we have used 2D CNN to recognize the human by their gait features by analyzing their walking pattern.
Research Paper: Accepted in the Journal of Physics and Nanotechnology.
Project 04
Model Evaluation Chapter Published in Book
Model Evaluation Chapter Published in Book
Model Evaluation (Chapter) in Scrivener Publishing
This is a chapter published in the book of Fundamentals and Methods of Machine and Deep Learning: Algorithms, Tools and Applications. In this chapter, I have explained how we can evaluate different kinds of model such as Computer Vision, NLP, Regression, Classification. and various other kind of model. This chapter discusses theoretical as well as mathematical steps to evaluate the trained model.
Chapter Link : Link
DOI (Book) : 10.1002/9781119821908
Project 05
Paper Published in Springer Journal
Paper Published in Springer Journal
Channel-Based Similarity Learning Using 2D Channel-Based Convolutional Neural Network
Research Paper named as Channel-Based Similarity Learning Using 2D Channel-Based Convolutional Neural Network published in Springer - Artificial Intelligence on Medical Data, Lecture Notes in Computational Vision and Biomechanics book series (LNCVB, volume 37)
Research Paper : Link