Developing an Algorithm for Scene Understanding and Decision Making Based on Visual Perception

摘要

Trending is the era of artificial intelligence, which now a day plays very crucial roles for performing daily repetitive as well as decision making complex tasks. Sensing the dynamic scenario and making decision based on visual perception is blooming a bit. In this we propose an algorithm in which we divide the image in two categories frame of reference and the visual objects. Frame of reference gives a platform on which the object is placed that deals in making environment decision. We use faster RCNN algorithm for object and background detection. In co-relation of object, there motion and environmental condition, for making decisions Naives bayes algorithm is developed which is used to sense the surrounding and automatically make a decision based on visual perception and coordinate needs. At the end we analysis the experimental results such as training accuracy, total losses function which has pretty fine reliability and accuracy.