Video Pedestrian Event Detection. Given a video and an event to be detected, identify, detect and track the people in the video, and make corresponding event judgments, such as smoking, helmets, gender, etc. 视频行人事件检测。给定一段视频和一个待检测事件，对视频中的人进行识别、检测和跟踪，并做出相应的事件判断，如吸烟、戴头盔、性别等。
DCSCN Super Resolution model in pytorch
This repository contains the codebase for an advanced video analysis pipeline that combines multiple state-of-the-art models and methodologies for high-accuracy activity recognition. The pipeline utilizes models and methodologies such as YOLO V8, ByteTrack, Movenet, and Transformer encoders.
An 8-class CNN built from scratch in PyTorch to classify blood cell types, achieving 93.8% accuracy on a hidden test set, with a systematic ablation study on BatchNorm and Dropout.
Real-time edge AI tracking for physical RC car bowling. Features a zero-allocation Android pipeline, Alpha-Beta-Gamma kinematic filtering, and YOLO26n inference on budget hardware.
Public shell for a spatial memory benchmark from egocentric video. Documentation-only: protocol, schema, and data card—no code or datasets.
Lightweight CNN for pressure ulcer staging (Stages I-IV) that runs on-device. 92% accuracy, outperforms DenseNet121 & MobileNetV2. Presidential AI Challenge state champion; ICIBM 2026.
AI-powered Android educational application for Odontogenic Oral Pathology using Java, Firebase, Google Gemini API, quiz generation, chatbot, AR/3D learning, and student performance analytics.
✨ My GitHub profile README — Meet Maru · AI & ML Engineer from Mumbai, India. Building production ML systems, AI agents & full-stack web apps. SIH participant · VP @ CSI VIVA · Open to internships.
Add a description, image, and links to the computer- topic page so that developers can more easily learn about it.
To associate your repository with the computer- topic, visit your repo's landing page and select "manage topics."