Export Reviews, Discussions, Author Feedback and Meta-Reviews

We sincerely thank all reviewers and AC for their time and effort. Our paper proposes a method for learning object proposals as a ConvNet that shares features/computation with a Fast R-CNN detection network, leading to real-time detection rates with state-of-the-art accuracy. We are glad that the reviews are consistently positive, and we appreciate the constructive suggestions for improving our paper. The reviewers noted that "the paper will have real practical impact in object detection" (R4), "the experimental results are solid" (R1, R5), and the paper "will be very interesting to the computer vision community as the results are impressive" (R6). We address the reviewer's concerns and questions as follows.