R-CNN LEARNING METHOD AND TESTING METHOD OF OBJECT DETECTOR TO BE USED FOR SURVEILLANCE BASED ON R-CNN CAPABLE OF CONVERTING MODES ACCORDING TO ASPECT RATIOS OR SCALES OF OBJECTS AND LEARNING DEVICE AND TESTING DEVICE USING THE SAME

摘要

The present invention relates to a method for learning an object detector based on a region-based convolutional neural network (R-CNN), wherein the aspect ratio and scale of an object including a traffic light are the distance from the object detector, the shape of the object, etc. It may be determined according to the same characteristic, the learning method comprising: a learning device, causing the RPN to generate an ROI candidate; Causing the pooling layer to output a feature vector; And learning the FC layer and the convolutional layer through backpropagation. In this method, the pooling process includes distance information and object information obtained through a radar, a lidar, or another sensor. It can be performed according to the actual ratio and the actual size of the object, and the learning method and the test method have a similar size from the same point of view at a specific location, so a method is provided that can be used for monitoring.