Human action classification system with pose-based (MediaPipe) and video-based (3D CNN) models. Features 100+ architectures for real-time pose classification and temporal models pretrained on UCF-101/HMDB51.

Securade.ai Sentinel - A monitoring and surveillance application that enables visual Q&A and video captioning for existing CCTV cameras.

Custom YOLO11m model for detecting and classifying car body damage (99% shattered glass, 96% flat tire detection accuracy)—optimized for high-capacity inference and assistive use in inspection and service workflows like BMW pre-loaner inspections.

A real-time inferencing of multistreaming YOWOv3(Spatio Temporal Action Detection task) using (UCF101-24) dataset. The repo is extension of https://github.com/Hope1337/YOWOv3, https://arxiv.org/pdf/2408.02623

YOLO-TLP: detected and classified tiny objects with bounding box dimensions smaller than 15 pixels, outperforming other one-stage detectors. maximum resolution for target observation in real-time applications.

This project uses YOLO for real-time leukemia detection in blood samples and CNNs for classifying brain hemorrhages in MRI scans. It aims to support faster, more accurate medical diagnostics through deep learning.

Explore a wide range of computer vision projects and documentation covering everything from object detection, image segmentation, and tracking to pose estimation, object counting, and automated annotation. These resources highlight real-world AI applications built with modern models like Ultralytics YOLO, Meta SAM 2, and other vlms

YOLOv12 Underwater Object Detection is an open-source suite for underwater object detection, built on YOLOv12. It offers an end-to-end pipeline with GPU-accelerated training, customizable data augmentations, real-time inference via Gradio, and support for model export (ONNX & PyTorch).

A real-time multi-person human pose estimation system using TensorFlow MoveNet Multipose (Lightning). Built with OpenCV for video and webcam inference, it detects and visualizes keypoints and skeletal connections with confidence-based filtering, optimized for speed and multi-person scenarios.

KeSSie HUGE Context Semantic recall for Large Language Models

This project uses YOLO for real-time leukemia detection in blood samples and CNNs for classifying brain hemorrhages in MRI scans. It aims to support faster, more accurate medical diagnostics through deep learning.

Improved PIDNet for real-time semantic segmentation. Work in progress.

THYX is an edge AI video analysis platform for AIoT, featuring behavior recognition, intelligent alerts, and real-time management. Modular, high-performance, and lightweight — ideal for smart security and industrial scenarios.

A conveyor belt sorting system powered by Raspberry Pi and YOLOv8 for real-time object detection.

Volleyball tracking - VballNet is a specialized deep learning framework designed for volleyball tracking, built upon the foundation of TrackNetV4. This repository includes two primary models, VballNetV1 and VballNetFastV1

40x faster AI inference: ONNX to TensorRT optimization with FP16/INT8 quantization, multi-GPU support, and deployment

Project Agora: MVP of the Concordia framework. An ethical, symbiotic AI designed to foster and protect human flourishing.

PlantAi is a ResNet-based CNN model trained on the PlantVillage dataset to classify plant leaf images as healthy or diseased. This repository includes PyTorch training code, tools to convert the model to TensorFlow Lite (TFLite) for deployment, and an Android app integrating the model for real-time leaf disease detection from camera images.

Bachelor Final Year Project exploring real-time speech denoising using machine learning. Compares classical methods (SS, WF, MMSE-LSA) with 5 deep models on spectrogram data, highlighting Conv-TasNet’s effectiveness. Features dataset bucketing, OOM mitigation, and batch evaluation.

This repository features an image sharpening pipeline using Knowledge Distillation. A high-capacity Restormer acts as the teacher model, while a lightweight Mini-UNet is trained as the student to mimic its performance.

AI-powered automated checkout system using YOLO11 object detection and face recognition to eliminate queues and reduce checkout times at retail facilities

A Spatial Retrieval-Augmented Generation system for latent world models, designed for embodied spatial intelligence in robotics, autonomous navigation, and embodied AI. Features ROS2 integration, real-time inference @ 25Hz, and complete robot build guide.

Deep voice speaker recognition system built with Keras CNNs. Educational project featuring audio augmentation, mel-spectrogram processing, and real-time inference. Binary classification for beginners.

Autonomous AI vision agent with contextual memory for real-time spatial computing | Multi-modal LLM reasoning + YOLO grounding + persistent AR overlays | Next-gen hands-free AI interface

Real-time Face Mask Detection using YOLOv8 — trained on custom dataset with live webcam inference.

Sovereign AI for smart, self-learning buildings • Edge-native • 25–35 % energy savings

Real time HSE compliance monitoring system using YOLOv8. Designed for Edge AI deployment to automate industrial safety protocols and personnel detection.

YOLOE-Unified is a novel framework that integrates YOLOE with distilled CLIP, runtime SAM refinement, and TensorRT optimization for efficient open-vocabulary object detection and instance segmentation on edge devices (Jetson Orin, etc.).

Custom object detection project using TensorFlow Lite Model Maker with an EfficientDet-Lite2 backbone. Trained to detect rocks and bags, deployed to Android for real-time inference on a Pixel 7a. Focused on efficient edge-device performance and streamlined model integration.

LLM-inspired BiLSTM pipeline for real-time, multi-label toxicity inference across adversarial discourse modalities.