Method

MambaGaze-Stereo: Enhanced Depth Estimation for Autonomous Driving via Selective Attention and State [MambaGaze-Stereo]
[Anonymous Submission]

Submitted on 1 Nov. 2024 13:41 by
[Anonymous Submission]

Running time:0.61 s
Environment:GPU @ 1.5 Ghz (Python)

Method Description:
My method introduces a MobileNetV4-based enhanced
feature extraction module in the IGEV model to
improve feature richness and disparity accuracy by
leveraging the efficient multi-scale extraction
capabilities of MobileNetV4. Additionally, the
FlashAttention-based Feature Correlation
Enhancement Module (FACEM) is proposed to enhance
feature correlation accuracy with a memory-
efficient attention mechanism, benefiting complex
scenes and occlusion handling. Finally, the Mamba
Selective Multi-Update Block (MSMUB) leverages
bidirectional selective attention within a multi-
scale structure, dynamically adjusting receptive
fields for fine-grained texture and edge detail
preservation, ensuring robust feature fusion
across resolutions for improved stereo matching
accuracy.
Parameters:
none
Latex Bibtex:
Not yet

Detailed Results

This page provides detailed results for the method(s) selected. For each of the first 20 test images, the number of erroneous pixels at all thresholds is depicted in the table. Underneath, the left input image, the disparity / end-point error map and the estimated (and interpolated) disparity / optical flow map are shown. The error map scales linearly between 0 (black) and >=5 (white) pixels error. Red denotes all occluded pixels, falling outside the image boundaries. The false color map is scaled to the largest ground truth disparity / flow value.

Test Set Average

This table as LaTeX

Reflective Regions

This table as LaTeX

Test Image 0

This table as LaTeX





Test Image 1

This table as LaTeX





Test Image 2

This table as LaTeX





Test Image 3

This table as LaTeX





Test Image 4

This table as LaTeX





Test Image 5

This table as LaTeX





Test Image 6

This table as LaTeX





Test Image 7

This table as LaTeX





Test Image 8

This table as LaTeX





Test Image 9

This table as LaTeX





Test Image 10

This table as LaTeX





Test Image 11

This table as LaTeX





Test Image 12

This table as LaTeX





Test Image 13

This table as LaTeX





Test Image 14

This table as LaTeX





Test Image 15

This table as LaTeX





Test Image 16

This table as LaTeX





Test Image 17

This table as LaTeX





Test Image 18

This table as LaTeX





Test Image 19

This table as LaTeX







eXTReMe Tracker