Multi-Scale Occ: 4th Place Solution for CVPR 2023 3D Occupancy Prediction Challenge
This is an incremental improvement for autonomous driving systems, focusing on enhancing 3D scene understanding in a specific competition setting.
The paper tackled the problem of 3D occupancy prediction by proposing Multi-Scale Occ, a method based on the lift-splat-shoot framework that uses multi-scale image features and temporal fusion, achieving 49.36 mIoU and ranking 4th in the CVPR 2023 challenge.
In this report, we present the 4th place solution for CVPR 2023 3D occupancy prediction challenge. We propose a simple method called Multi-Scale Occ for occupancy prediction based on lift-splat-shoot framework, which introduces multi-scale image features for generating better multi-scale 3D voxel features with temporal fusion of multiple past frames. Post-processing including model ensemble, test-time augmentation, and class-wise thresh are adopted to further boost the final performance. As shown on the leaderboard, our proposed occupancy prediction method ranks the 4th place with 49.36 mIoU.