Recovering 6D Object Pose: Multi-modal Analyses on Challenges
share
A large number of studies analyse object detection and pose estimation at visual level in 2D, discussing the effects of challenges such as occlusion, clutter, texture, etc., on the performances of the methods, which work in the context of RGB modality. Interpreting the depth data, the study in this paper presents thorough multi-modal analyses. It discusses the above-mentioned challenges for full 6D object pose estimation in RGB-D images comparing the performances of several 6D detectors in order to answer the following questions: What is the current position of the computer vision community for maintaining "automation" in robotic manipulation? What next steps should the community take for improving "autonomy" in robotics while handling objects? Direct comparison of the detectors is difficult, since they are tested on multiple datasets with different characteristics and are evaluated using widely varying evaluation protocols. To deal with these issues, we follow a threefold strategy: five representative object datasets, mainly differing from the point of challenges that they involve, are collected. Then, two classes of detectors are tested on the collected datasets. Lastly, the baselines' performances are evaluated using two different evaluation metrics under uniform scoring criteria. Regarding the experiments conducted, we analyse our observations on the baselines along with the challenges involved in the interested datasets, and we suggest a number of insights for the next steps to be taken, for improving the autonomy in robotics.
READ FULL TEXT