Kanchan Sarkar's repositories
Highly-Optimized-Interactive-Image-Segmentation
This is a repo for highly optimised interactive image segmentation model. This is a Tensorflow implementation of Interactive Image Segmentation with Latent Diversity. Using UI we collect positive and negative clicks provided by user and perform the interactive segmentation.
system-design-primer
Learn how to design large-scale systems. Prep for the system design interview. Includes Anki flashcards.
Facebook-Profile-Based-TV-Recommender-System
We implement and evaluate several algorithms in the context of developing a recommender system based on data gathered from Facebook user profiles. In particular, we look at a Matrix Factorization technique (SVD), a Clustering algorithm (K-Means), two Collaborative Filtering algorithms, a Content-Filtering approach, Latent Semantic Analysis (LSA), Link Prediction, and Na¨ıve Bayes, and compare their performance in terms of standard measures. The algorithms draw from principles and techniques in Machine Learning, Natural Language Processing, Information Retrieval, as well as Graph Theory.
Salient-Object-Detection
This is the PyTorch implementation of paper 'DenseFNet: Dense Encoder-Decoder Network with Focal Loss for Salience Object Detection'.
coding-interview-university
A complete computer science study plan to become a software engineer.
Optimization-Object-Detection-Model
Most of the SOTA object detection model are too heavy to deploy in production for real time application. We optimize the most popular one stage detector YOLOV3 for deployment on NVIDIA Jetson Nano Developer Kit. Reduces model size by 40% and throughput increased by 2X without compromising the accuracy.
ExData_Plotting1
Plotting Assignment 1 for Exploratory Data Analysis
MultiMarkdown
Expanded perl version of John Gruber's original Markdown --- No longer under active development since MMD 3
yolov3-tiny-onnx-TensorRT
convert your yolov3-tiny model to trt model