Parth Vasoya's repositories
Carpool_Management_System
Developed a Carpool Management System using Node.js for the core trip execution engine. Implemented a data visualization tool using react-map-gl for monitoring.
High-Conflict-Driving-Scenario-Negotiation
Implemented multi-agent reinforcement learning (MARL) approach for robust negotiation of Bi-directional Lane Usage with Discrete Asymmetric Soft Actor-Critic (DASAC) algorithm.
Carpool-graph
Efficient Carpool route algorithm
Drowsiness-Detection
A Computer Vision-based system to detect driver drowsiness by detecting facial features like eyes and mouth.
Digital_Image_Watermaking
Deployed a digital watermarking scheme based on SVD, DWT, and DCST with improved bandwidth efficiency and robustness (PSNR 26.63) in Heroku using flask.
Lane-Detection-in-Unstructured-Environments
A novel recurrent neural network for lane detection based on LSTM. Achieved 2% improvement in accuracy and 150% reduction in inference cost on Indian Driving Dataset (IDD) over CRF-based methods
R-fresh-COVID-19
An Android application to facilitate online ordering of grocery items and eatables from a local grocery store during the COVID-19 pandemic.
Real-time_Drowsiness_Detection
A CNN-based drowsiness detector with Dlib facial feature extractor. Designed a sub-network based on lightweight architectures for yawn detection. Implemented a CNN based on attention mechanism for eye gaze estimation
Unknown-Road-Objects-Perception
Designed a CNN based on a category-agnostic instance embedding concept for identifying unknown road objects using point clouds from a BEV LiDAR frame.
YOLOv3_Object_Tracking
Optimized a YOLOv3-Deep SORT object tracker through Dlib integration and minimized loss to 4.73%
Chatbot_Heroku
Build a chatbot using Heroku!
Lane_Detection_IDD_Dataset
Lane detection using CNN with separate embedding and segmentation branches that yield embeddings on each lane pixel. Adopted instance segmentation approach to alleviate the issue of changing lane counts.
YOLOv3_Traffic_Sign_Detection
A YOLOv3 object detector to identify people and traffic signs on OpenImageV6 1.6M images
Deep_Sort
Online Realtime Tracking with YOLOv3
Face-Mask-Detection
A CNN architecture that employs multi-task learning to detect the presence of a face mask and its coverage of the nose, mouth, and chin on human faces. Conducted training on MaskedFace-Net 138k synthetic images.
Parth_Vasoya
https://parthkvv.github.io/Parth_Vasoya/
pytorch-image-models
PyTorch image models, scripts, pretrained weights -- ResNet, ResNeXT, EfficientNet, EfficientNetV2, NFNet, Vision Transformer, MixNet, MobileNet-V3/V2, RegNet, DPN, CSPNet, and more
RetinaNet_Object_Detection
Object detection using RetinaNet, a single stage object detector, for performance comparison with YOLO.
Robust-Visual-Place-Recognition
A place recognition algorithm with CNN feature extraction for scene matching and description to handle perception, appearance changes.
ROS_Bot_Navigation
ROS bot navigation with lane-line detection.
sqlite-utils
Python CLI utility and library for manipulating SQLite databases
starting-react
My starting react repository(sample project)
TaskSync-Pro
A task management system, inspired by Asana, utilizing React.js for the front end, Node.js and Express.js for backend operations, WebSocket for real-time updates, and MongoDB for secure data management.
TensorFlow-2.x-YOLOv3
YOLOv3 implementation in TensorFlow 2.3.1
VisualPhrase
VisualPhrase: Insightful Image Captioning
Work-Together
A team collaboration website for efficiently carrying out professional tasks within an organisation, implementing the concept of portfolios, kanbans and timelines.
YOLOv6_Unstructured_Condition
YOLOv6 implementation on Indian Roads
YOLOv7_Unstructured_Evaluation
YOLOv7 implementation on Indian Roads