A Conceptual Investigation on the Image Processing using Artificial Intelligence and Tensor Flow Models through Correlation Analysis

Artificial intelligence plays a vital role in image processing and tensor flow models. Due to increasing demand of generating data from image as well as to make distinction between images the necessity and relevance of these two topics are also growing. Image classification and processing is a recent trend where traffic lights and other objects are identified by Machine Learning (ML) and Deep Learning (DL) technologies. These technologies are further implemented inside self-driven vehicles for developing autonomous processing's and information transfer. However, an accuracy and speed of detection process is a concern in today's world. Thus, this study paper provides an insight into image processing accuracy by “single-stage object detection model” or SSD MobileNetV2 and “Regional-based Convolutional Neural Network” or RCNN. Pearson correlation analysis and mean was obtained using the input data. The output data were further analysed and discussed. Findings suggested that RCNN is 80.90% accurate when it is allowed to detect traffic lights present at 11.9 metres apart; whereas SSD MobileNet V2 has shown 71.8% of accuracy when the traffic light was detected present at the same distance. On average, the RCNN shows higher accuracy than SSD MobileNet V2. The discussion found that RCNN is more accurate and slower than SSD MobileNetV2.