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 $d_{i}$ and produces a confidence map, which can be thresholded to yield the binary segmentation. To estimate depth on $N + 1$ quantization levels we run this network $N$ 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.
Source: Bi3D: Stereo Depth Estimation via Binary ClassificationsPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
3D Object Detection | 1 | 11.11% |
Active Learning | 1 | 11.11% |
Domain Adaptation | 1 | 11.11% |
Object Detection | 1 | 11.11% |
Unsupervised Domain Adaptation | 1 | 11.11% |
Autonomous Navigation | 1 | 11.11% |
Depth Estimation | 1 | 11.11% |
Quantization | 1 | 11.11% |
Stereo Depth Estimation | 1 | 11.11% |
Component | Type |
|
---|---|---|
🤖 No Components Found | You can add them if they exist; e.g. Mask R-CNN uses RoIAlign |