Viet-Anh on Software Logo

What is: Bi3D?

SourceBi3D: Stereo Depth Estimation via Binary Classifications
Year2000
Data SourceCC BY-SA - https://paperswithcode.com

Bi3D is a stereo depth estimation framework that estimates depth via a series of binary classifications. Rather than testing if objects are at a particular depth D, as existing stereo methods do, it classifies them as being closer or farther than D. It takes the stereo pair and a disparity d_id\_{i} and produces a confidence map, which can be thresholded to yield the binary segmentation. To estimate depth on N+1N + 1 quantization levels we run this network NN times and maximize the probability in Equation 8 (see paper). To estimate continuous depth, whether full or selective, we run the SegNet block of Bi3DNet for each disparity level and work directly on the confidence volume.