#computer-vision

Verifiable handwriting synthesis for failure-driven multimodal training and regression.

Eyes for AI coding agents 👁️ — render → perceive → report → fix → re-render. A machine-graded visual feedback loop (DOM/contrast/OCR-grounded + optional vision LLM) agents consume to self-correct before claiming done.

AI-powered Smart Farming Rover for Plant Disease Detection, Autonomous Navigation, and Real-time IoT Monitoring

AI-powered event photo delivery platform — guests find and download their own photos from thousands of event images using face recognition, no manual sorting required. Full-stack: React, Node/Express, Python/FastAPI, MongoDB, Cloudinary.

Local-first AI cognitive operating system with memory, vision, voice, PDF intelligence, CCTV analysis, temporal replay, and agent workflows.

AI-powered attendance management system using Flask, OpenCV, and face_recognition with student registration and teacher dashboard.

Private, offline multimodal AI for Android using Qwen3-VL, llama.rn, and Whisper.

AI voice assistant for blind and low-vision users — point a phone camera, ask a question out loud, and get a concise spoken answer. FastAPI + Flutter, with Egyptian-Arabic TTS, currency detection, barcode/product lookup, and object-finding.

Evidence-first industrial vision data governance Agent for GOAI Boundless Agents

AI-powered plant disease detection web app. Upload leaf photos for real-time diagnosis using GPT-4 Vision or Claude 3 Opus. Get detailed reports with severity, symptoms, treatment, and prevention. Built with React, FastAPI, and Tailwind CSS.

The copilot windows never gave you, always one keystroke away Press Alt+P. A silent command bar appears over whatever you're doing. Type. Your result lands in the clipboard, streams into a chat, becomes a terminal agent, a background task or a computer use agent. Then it vanishes.

Evidence-backed engineering portfolio focused on computer vision, deep learning, machine learning, computer networks, and digital and wireless communications.

Generate handwriting images from your own handwriting — Python text-to-handwriting generator, handwriting synthesis. Build a glyph library, then render text in your real strokes. No font rendering, Chinese, offline, no AI model, without AI. 用你自己的手写字生成手写风图片：手写生成、仿手写，从手写照片建字形库，用本人真实笔画渲染文字，非字体渲染、离线、不用 AI 模型。Local-first, privacy-preserving.

Real-time object detection pipeline · YOLO + ONNX · DXGI capture · low-latency C#/.NET 8 research project

Identify bird species from images along with audio and explore them with AI-powered insights.

AI-powered skin lesion classifier using ResNet18 transfer learning on HAM10000 dataset — PyTorch + Streamlit

Ranked AI/ML conference and journal tracker with submission deadlines, tentative annual windows, source links, and update status.

INTERVION is a fully local, real-time AI interview simulator built for developers and students preparing for technical interviews.

Deep learning-based web application for automated diabetic retinopathy detection from retinal fundus images using ResNet50 and TensorFlow.

Production-oriented AI inspection platform combining Computer Vision, LLMs, FastAPI, and Docker to generate automated engineering reports from industrial images.

AI-powered fitness assistant for workout tracking, nutrition guidance, habit tracking, chatbot support, and personalized fitness recommendations.

Evidence-based technical case studies on reliable LLM applications, Agent systems, RAG, and AI backend engineering.

🌿 End-to-end full-stack AI Crop Disease Prediction & Agronomist Platform with 67 classes across 22 crops, Grad-CAM explainability, weather risk forecasting, and AI Chatbot.

101 Günlük Yapay Zeka, Bilgisayarlı Görü, LLM/RAG ve MLOps Mühendisliği Master Programı — MiniViT-MoE v2, FastAPI, Docker ve Uçtan Uca AI Sistemleri

Enterprise AI-powered drone infrastructure inspection platform with Google ADK, Gemini, Mission Planner, MAVSDK, Digital Twin, Vision AI, and Predictive Maintenance.

End-to-end Deep Learning & Explainable AI (XAI) dermatological diagnostic web app using YOLOv8-seg (0.91 mAP@50), MobileNetV2 (91.76% accuracy), Grad-CAM, RAG chatbot, and Flask.

AI-powered cell nuclei segmentation and counting web platform

Real-time AI fitness platform using MediaPipe, OpenCV, Flask and React with pose estimation and workout analytics.

Touchless AI Music Experience using Computer Vision

Real-time browser air-writing board: pinch-to-draw on a virtual whiteboard over your webcam feed using MediaPipe hand landmark tracking.

Play guitar chords in the air with your webcam, MediaPipe hand tracking, Karplus–Strong synthesis, no samples on disk

Execution API that gives language models (Claude Code, Gemini…) the ability to see web interfaces — Playwright · Set-of-Mark · ReAct

Autonomous AI-powered hospital supply delivery robot using ROS, YOLOv8 edge vision, and real-time cloud telemetry.

AI-powered meeting intelligence platform with real-time video conferencing, focus analytics, attendance tracking, AI-generated meeting insights, and PDF reporting.

Gaze + voice as a multimodal input layer — look, speak, and your gaze-resolved intent goes to Claude.

Autonomous Research with AI: can a UAV learn a city by heart? An AI agent running an autoresearch loop (81 experiments, $364) found a 3.1 MB neural memory that geolocates in-flight pictures of Berlin to ~27 m — no GPS, no map on board, built for a <$5 ESP32-P4 flight computer.

A hosted web app that photographs a book cover, extracts title/author/metadata using Claude's vision API, and logs it to a Notion library — built as a reusable "photo → structured data" engine designed to extend beyond books (e.g. small business inventory scanning). Replaces an earlier iPhone Shortcut prototype with a branded, always-on service.

Sidelook, a Practical Systems product. A Windows desktop companion: ask about the window in front of you, build a prototype from a sketch, work an app one approved click at a time, on the ChatGPT or Claude subscription you already pay for.

Anivar NVR — a free, open-source Blue Iris alternative with on-device AI: faces, plates, sounds and plain-language search. A desktop app, no Docker.

AI interview prep & performance analyzer — generates CV-grounded STAR interview guides via RAG, then scores practice video answers on content, speech, and visual delivery using Whisper, librosa & MediaPipe.