A Region-Based Convolutional Neural Network (R-CNN) is a family of object detection models that first generate region proposals and then classify each proposed region using CNN features​.In this two-stage approach, an algorithm (such as a Region Proposal Network or selective search) proposes candidate bounding boxes in an image that might contain objects. Each candidate region is then fed through a convolutional neural network to extract features, and a classifier predicts the object class for that region (and often refines the bounding box). R-CNN and its successors (Fast R-CNN, Faster R-CNN) effectively handle multiple objects and scales, achieving accurate object localization and classification​.They are robust to varying object sizes, occlusions, and backgrounds by focusing on pertinent regions of interest in the image.​