This project uses YOLOv11-nano to detect and classify solar panel conditions into six categories: Clean, Dust, Bird, Electrical, Physical, and Snow. The goal is to automate fault detection, reduce manual inspection time, and help maintain optimal solar panel efficiency using computer vision and deep learning.

A prototype that detects speed signboards (20, 40, 60, STOP) using a custom-trained YOLOv11 model and automatically adjusts the movement of a small car using motor control. Designed for real-world scenarios like hospital, school, and highway zones.

Official implementation of An Explainable AI based Plant Disease Identification using a Two-Stage Detection-Classification Pipeline with YOLO and ECA-NFNet Framework, 28th International Conference on Computer and Information Technology (ICCIT, 2025)

This repository demonstrates how to use a pretrained YOLOv11 model from the Ultralytics library for image-based object detection.

Web Application that provides real-time American Sign Language to English (Text & Voice) translation.

A real-time object detection mobile application built with Flutter, integrating YOLOv8n on the COCO dataset via TensorFlow Lite for accurate, on-device detection with adjustable confidence thresholds.

visionhub-ai is a Django-based application that demonstrates a powerful pipeline combining YOLOv11n-seg (instance segmentation) with bird classification and face recognition features.

Yolov11n model benchmarking on 20000+ fire images across multiple epochs.

This repository is a structured collection of my Computer Vision learnings, implementations, and experiments from my first year in industry. It reflects real-world challenges, solutions, and best practices I encountered while building and deploying vision-based systems.

Geometriai adatelemzés nevű tárgyra készített projektmunkám

This project presents a hybrid deep learning framework for automatic license plate recognition (ALPR), specifically tailored for Addis Ababa minibus taxis.

AI camera survillance

🔍 Use a pretrained YOLOv11 model for efficient image object detection, enabling quick comparison of original images and model outputs.