3D human reconstruction from RGB images achieves decent results in good weather conditions but degrades dramatically in rough weather. To address this, mmWave radars have been employed to reconstruct 3D human joints and meshes in rough weather. However, combining RGB and mmWave signals for weather-robust 3D human reconstruction is challenging due to the sparse nature of mmWave and the vulnerability of RGB images. To overcome these challenges, a large-scale mmWave dataset was created, consisting of synchronized and calibrated mmWave radar point clouds and RGB(D) images under different weather conditions. Based on this dataset, ImmFusion was designed as the first mmWave-RGB fusion solution to robustly reconstruct 3D human bodies in various weather conditions. ImmFusion consists of image and point backbones for token feature extraction and a Transformer module for token fusion. The image and point backbones refine global and local features from original data, and the Fusion Transformer Module aims for effective information fusion of two modalities by dynamically selecting informative tokens. Extensive experiments demonstrate that ImmFusion can efficiently utilize the information of two modalities to achieve robust 3D human body reconstruction in various weather environments.
Transformer
Image and point backbones, Fusion Transformer Module
Large-scale mmWave dataset
Robustness in various weather conditions
Not specified
No
No
Weather-robust, mmWave-RGB fusion, Transformer-based
No
Not specified
Not specified
Not specified
Not specified
Not specified
Not specified
No
Not specified
Not specified
Not specified
Not specified
Not specified
Not specified
Not specified
Not specified
Not specified
3D human body reconstruction
Not specified
Not specified
Not specified
Not specified
Not specified
Not specified
No
Not specified
Not specified
No
Not specified
Not specified
Not specified
Not specified
Not specified
No
Not specified
Not specified
0.00
Not specified
Not specified
01/01/1970
01/01/1970
Not specified
Not specified
Yes