Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks

Abstract

State-of-the-art object detection networks depend on region proposal algorithms to hypothesize object locations. Advances like SPPnet [1] and Fast R-CNN [2] have reduced the running time of these detection networks, exposing region proposal computation as a bottleneck. In this work, we introduce a Region Proposal Network (RPN) that shares full-image convolutional features with the detection network, thus enabling nearly cost-free region proposals. An RPN is a fully convolutional network that simultaneously predicts object bounds and objectness scores at each position. The RPN is trained end-to-end to generate high-quality region proposals, which are used by Fast R-CNN for detection. We further merge RPN and Fast R-CNN into a single network by sharing their convolutional features-using the recently popular terminology of neural networks with 'attention' mechanisms, the RPN component tells the unified network where to look. For the very deep VGG-16 model [3] , our detection system has a frame rate of 5 fps (including all steps) on a GPU, while achieving state-of-the-art object detection accuracy on PASCAL VOC 2007, 2012, and MS COCO datasets with only 300 proposals per image. In ILSVRC and COCO 2015 competitions, Faster R-CNN and RPN are the foundations of the 1st-place winning entries in several tracks. Code has been made publicly available.

DeepChess: End-to-End Deep Neural Network for Automatic Learning in Chess

“…It is trained to predict which of the two positions results in a win. Note that similar to the most successful object detection methods [11], we found a 2-value output to outperform one binary output. Figure 1 illustrates the neural network architecture.…”

DAPs: Deep Action Proposals for Action Understanding

“…achieving high object recall with less number of bounding boxes, preferably with a small computational overhead and the potential to scale to hundreds of object categories [35,40,37]. Here, discriminative methods based on deep learning models have helped improve the ranking quality of proposal approaches [7,37,28,32]. Inspired by this work, we extend the use of deep and recurrent networks to temporal action proposal generation by introducing a new architecture.…”

“…of efficiency and high detection rates [29,28]. Efficient object proposal modules have also enabled a boost in performance of other high-level visual tasks, such as simultaneous detection and segmentation, object tracking, and image captioning [13,14,16,20].…”

Current trends in the development of intelligent unmanned autonomous systems

“…In vision, abstractions can include object detection (Girshick et al, 2014;Girshick, 2015;Ren et al, 2015), classification (Krizhevsky et al, 2012;Simonyan and Zisserman, 2014;Szegedy et al, 2015), and semantic understanding (Huang et al, 2013) using convolution neural networks (LeCun and Bengio, 1995). Inspired by the hierarchical architecture of the human visual cortex (Hubel and Wiesel, 1962), architectures for multiple convolution-pooling layers have been proposed and are being used in different machine learning tasks.…”