A Robotic Visual Grasping Design: Rethinking Convolution Neural Network with High-Resolutions. (arXiv:2209.07459v2 [cs.RO] UPDATED)

High-resolution representations are important for vision-based robotic
grasping problems. Existing works generally encode the input images into
low-resolution representations via sub-networks and then recover
high-resolution representations. This will lose spatial information, and errors
introduced by the decoder will be more serious when multiple types of objects
are considered or objects are far away from the camera. To address these
issues, we revisit the design paradigm of CNN for robotic perception tasks. We
demonstrate that using parallel branches as opposed to …

arxiv convolution convolution neural network design network neural network