Novel approach for object detection neural network model using opencv and mask RCNN algorithm accuracy and latency will be compared with SSD algorithm

The aim of this proposed work is to evaluate the performance of Single shot multibox detector(SSD) algorithms in identification of objects in image by comparing it with MASK Regional convolutional neural network(RCNN) algorithm. Comparison was based on their accuracy and latency.: A total of 100 samples images are collected from various classes and labels. These samples are divided into training dataset (60%) and test dataset (40%) Accuracy, Latency values were calculated to quantify the performance of the SSD and MASK RCNN with G power 0.8. The latency in prediction of the object in the image was lower in SSD(single shot multibox detector) 84.2354 millisecond compared to MASK RCNN algorithm 18.4656 sec. The sigma value was 0.124. It was proven that the SSD perform slightly better in the terms of latency.