Abstract:
The image pixels are severely distorted which is then caused by environmental conditions, compression techniques, and hardware restrictions. This results in poor quality images which affect several disciplines such as medical imaging, security surveillance, and digital media. This problem is tackled by creating an AI-based system that can identify and classify noise in real time, this system targets Noise, AWGN, jpeg compression, contrast, and blur. The image distortions are classified with a fast and precise deep learning model MobileNet V2. For real time monitoring, the machine uses Jetson Nano, an edge AI device. This deployment method eliminates the need for cloud computing which improves the latency and lowers the power usage. The AI allows for more reliable image quality assessment therefore it is very efcient for tasks that require high precision. Some potential focuses for the advanced development include improvement in distortion detection and modification of the real-time performance.