Ajinkya Pahinkar
Data Science Graduate
ABOUT ME
RESUME
PROJECTS
PUBLICATIONS
CONTACT
More
Faster Region Based Convolutional Neural Network and VGG 16 for Multi-Class Tyre Defect Detection presented at ICCNT
This paper intends to create a deep-learning architecture for training models over various defective tyre images until it has achieved intelligence to distinctly classify multi-class tyre defects and tell apart various types of defects in tires using computer vision. This paper aims to create a system that can detect defects and allow professionals to spend more time trying to find a fix for it instead of spending time diagnosing it.
Multi-Contrast Convolution Neural Network and Fast Feature Embedding for Multi-Class Tyre Defect Detection presented at ICECA
An automated tyre defect detection system created using state of the art deep learning technique of Convolutional Neural Networks. The two algorithms constructed using a baseline Resnet architecture includes - Multi-Contrast Convolution Neural Network and Fast Feature Embedding. This paper solves a multi class classification problems that can be used to tell apart different type of tyre defects such are tread wear indicator, bulge, linear air, sidewall crack and normal tyre images.
A Complete Guide to Using TensorBoard with PyTorch
A guide to using TensorBoard, a popular supporting tool of TensorFlow to create plots for training and accuracy of deep learning models. It comprehensively covers the installation and usage of TensorBoard with PyTorch to create loss, accuracy plots, histogram and distribution plots of the weights along with hyperparameter tuning techniques
Implementing CNN in PyTorch with Custom Dataset and Transfer Learning
A coding tutorial on working with convolutional neural networks with the help of a popular deep learning framework - PyTorch. The guide covers creating custom datasets with torch data loader functions, creating pre trained models with torch vision library and creating training loops in PyTorch.