Dr. Arun Singh Pundir

Dr. ARUN SINGH PUNDIR- Education- PhD, Indian Institute of Technology Roorkee, IIT Roorkee. Computer Science and Engineering - Masters of Technology, Kurukshetra University, Kurukshetra. Computer Science and Engineering - Bachelor of Technology, College of Engineering Roorkee, Roorkee. Computer Science and Engineering PhD Work Thesis Title Multi-Color-Model Video Analysis For Fire And Smoke Detection Supervisor Prof. Balasubramanian Raman, CSE Department, IIT Roorkee Domain Machine Learning and Deep Learning Relevant Projects (2015 - Present) 1 Classification of smoke and non-smoke frames using transfer learning where fine tuning was done on not only frames but also on optical flow extracted from smoke and non-smoke videos (Using deep learning and machine learning concepts) 2 Implemented ResNet and 100 layer DenseNet on CIFAR10 dataset from scratch using Keras API in TensorFlow 2.0 and python 3.6 3 Automatic colorization using deep-autoencoder for gray scale images using TF 2.0 on CIFAR10 dataset 4 Implemented Conditional GAN using Keras API and TF 2.0 on MNIST Dataset 5 Visual Search Using Transfer Learning (backbone networks VGG16, ResNet50 and InceptionV3 for feature extraction) - further using the concept of ImageDataGenerator, saved a model using checkpoints and plot results using Matplotlib 6 Style transfer- Implemented using VGG16 as a feature extractor utilizing TF 2.0 7 Used a CNN, RNN, LSTM and GRU for sentiment analysis using Kaggle dataset 8 Trained pretained YOLO v3 on custom dataset 9 Gender classification based on speech using deep learning (AlexNet and VGG on mozilla dataset having 60000 labelled voice samples) 10 Deep learning model for forest fire detection using transfer learning (VGG) and optical acceleration B- 1/3 Research Interests { Machine Learning { Computer Vision { Deep Learning { Object detection and segmentation { Convolutional Neural Network (Con- { Generative vNet/CNN) (GANs) Adversarial Networks Language and Tools Used { { { { { Tensorflow 2.0 { OpenCV { matplotlib for visualization { C language Python numpy Keras API scikit-learn Work Experience Jan. 2019– Assistant Professor, National Institute Of Technology Delhi, Delhi (With June as Dec. 2019 summer break). Aug. 2013– Assistant Professor, Uttaranchal Institute Of Technology, Dehradun. Dec. 2013 Feb. 2011– Lecturer, Uttaranchal Institute Of Technology, Dehradun. June 2011 Jan 2010– Lecturer, College Of Engineering Roorkee, Roorkee. Oct. 2010 Awards and Scholarships- MHRD Scholarship for PhD. 2017 Microsoft research grant to present paper at TENCON 2017 held in Malaysia (5-8 November 2017)- MHRD-GATE Scholarship for M.Tech course. Journal Publications 1 A. S. Pundir, B. Raman, Dual Deep Learning Model for Image Based Smoke Detection, Fire Technology (Springer) 55 (6) - A. S. Pundir, B. Raman, Deep belief network for smoke detection, Fire Technology (Springer) 53 (6) - International Conferences 1 A. S. Pundir and B. Raman, Gender classification based on speech deep learning, (Submission for 6th International Conference on Computer Vision and Image Processing CVIP 2021) 2 A. S. Pundir and B. Raman, Deep learning model for forest fire detection using transfer learning and optical acceleration, (Submission for 6th International Conference on Computer Vision and Image Processing CVIP 2021) B- 2/3 3 R. Bansal, A. S. Pundir and B. Raman, Dynamic texture using deep learning, 13th IEEE Region 10 Conference (TENCON 2017), pp-, November 5-8, 2017, Penang, MALAYSIA (Awarded the Microsoft research grant to present this paper) 4 A. S. Pundir, H. Buckchash, A. S. Rajput, V. K. Tanwar and B. Raman, Fire detection using dense trajectories, In Proceedings of 2nd International Conference on Computer Vision and Image Processing, (CVIP WM-2017), September 10-12, 2017, Greater Noida, INDIA References 1 Dr. Balasubramanian Raman, Professor (Computer Science and Engineering), IIT Roorkee, -,-2 Dr. Dharmendra Singh, Professor (HoD Computer Science and Engineering), IIT Roorkee, -,- B- 3/3