Deep Active Perception for Object Detection using Navigation Proposals
This work addresses the problem of improving object detection in robotics by enabling active perception with existing detectors, though it is incremental as it builds on prior active perception methods.
The paper tackles the limitation of static inference in deep learning for object detection by proposing a supervised active perception pipeline that uses an additional neural network to estimate better viewpoints when detector confidence is low, and it demonstrates effectiveness on synthetic datasets in the Webots simulator.
Deep Learning (DL) has brought significant advances to robotics vision tasks. However, most existing DL methods have a major shortcoming, they rely on a static inference paradigm inherent in traditional computer vision pipelines. On the other hand, recent studies have found that active perception improves the perception abilities of various models by going beyond these static paradigms. Despite the significant potential of active perception, it poses several challenges, primarily involving significant changes in training pipelines for deep learning models. To overcome these limitations, in this work, we propose a generic supervised active perception pipeline for object detection that can be trained using existing off-the-shelf object detectors, while also leveraging advances in simulation environments. To this end, the proposed method employs an additional neural network architecture that estimates better viewpoints in cases where the object detector confidence is insufficient. The proposed method was evaluated on synthetic datasets, constructed within the Webots robotics simulator, showcasing its effectiveness in two object detection cases.