Understanding Particles From Video
Property Estimation of Granular Materials via Visuo-Haptic Learning
IEEE Robotics and Automation Letters (2024), with ICRA 205.
Zeqing Zhang, Guangze Zheng, Xuebo Ji, Guanqi Chen, Ruixing Jia, Wentao Chen, Guanhua Chen, Liangjun Zhang and Jia Pan
Overview
In this paper, we introduce a method for estimating the relative values of particle size and density from the video of the interaction with granular materials (GMs).
Particle Tracking Algorithm
Setting for particle tracking algorithm on the given video. (a) We outline the ROI at the beginning frame, i.e., 120-th frame, and discretize it as 27 x 24 points. These points will be attached to granules and will follow granules in the remaining frames. (b) For every 4 frames, we will update a row of points at the bottom of ROI to be tracked in the subsequent streaming.
Particle tracking performance on the dataset GM15-VF, hand-collected data, and data collected on the beach.
Comparison Method: Image-Force Method (IF)
Some example inputs of the IF method are displayed as follows.
We extract the edge in RGB images about GMs with normalized pixel values in
[-1,1], which is improved from a deep visuo-tactile learning work [16].