Improving CNN-based Planar Object Detection with Geometric Prior Knowledge

Abstract

In this paper, we focus on the question: how might mobile robots take advantage of affordable RGB-D sensors for object detection? Although current CNN-based object detectors have achieved impressive results, there are three main drawbacks for practical usage on mobile robots: 1) It is hard and time-consuming to collect and annotate large-scale training sets.2) It usually needs a long training time. 3) CNN-based object detection shows significant weakness in predicting location. We propose a novel approach for the detection of planar objects, which rectifies images with geometric information to compensate for the perspective distortion before feeding it to the CNN detector module, typically a CNN-based detector like YOLO or MASK RCNN. By dealing with the perspective distortion in advance, we eliminate the need for the CNN detector to learn that. Experiments show that this approach significantly boosts the detection performance. Besides, it effectively reduces the number of training images required. In addition to the novel detection framework proposed, we also release an RGB-D dataset for hazmat sign detection. To the best of our knowledge, this is the first public-available hazmat sign detection dataset with RGB-D sensors.

DeepHAZMAT: Hazardous Materials Sign Detection and Segmentation with Restricted Computational Resources

“…The trained model is also able to detect occluded, overlapped, and partially visible signs. The experimental results showed that the DeepHAZMAT model is more accurate and faster than many other recent and state-or-the-art research works such as [18], [19], and [20]. The developed DNN-based system was fast enough to be implemented in Mobile robots, using a single Intel NUC Corei7 embedded system for robust and real-time hazard label detection, recognition, identification, localisation, and segmentation, thanks to skipping redundant input frames as well as adaptation of the YOLOv3-tiny for our real-time robotics application.…”