Abstract:3D human reconstruction from RGB images achieves decent results in good weather conditions but degrades dramatically in rough weather. Complementarily, 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 still an open challenge, given the sparse nature of mmWave and the vulnerability of RGB images. The limited research about the impact of missing points and sparsity features of mmWave data on reconstruction performance, as well as the lack of available datasets for paired mmWave-RGB data, further complicates the process of fusing the two modalities. To fill these gaps, we build up an automatic 3D body annotation system with multiple sensors to collect a large-scale mmWave dataset. The dataset consists of synchronized and calibrated mmWave radar point clouds and RGB(D) images under different weather conditions and skeleton/mesh annotations for humans in these scenes. With this dataset, we conduct a comprehensive analysis about the limitations of single-modality reconstruction and the impact of missing points and sparsity on the reconstruction performance. Based on the guidance of this analysis, we design ImmFusion, the first mmWave-RGB fusion solution to robustly reconstruct 3D human bodies in various weather conditions. Specifically, our 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. In addition, our method achieves superior accuracy compared to that of the state-of-the-art Transformer-based LiDAR-camera fusion methods.

Towards Weather-Robust 3D Human Body Reconstruction: Millimeter-Wave Radar-Based Dataset, Benchmark, and Multi-Modal Fusion

ImmFusion: Robust mmWave-RGB Fusion for 3D Human Body Reconstruction in All Weather Conditions

Mmskeleton: 3D Human Skeleton Estimation Using Millimeter Wave Radar Sparse Point Clouds

mmBody Benchmark: 3D Body Reconstruction Dataset and Analysis for Millimeter Wave Radar

Mpose: Environment- and Subject-Agnostic 3D Skeleton Posture Reconstruction Leveraging a Single Mmwave Device

AdaptiveFusion: Adaptive Multi-Modal Multi-View Fusion for 3D Human Body Reconstruction

MMBaT: A Multi-task Framework for mmWave-based Human Body Reconstruction and Translation Prediction

MI-Mesh: 3D Human Mesh Construction by Fusing Image and Millimeter Wave

MI-Mesh

Radar and Camera Fusion for Multi-Task Sensing in Autonomous Driving

All-weather Multi-Modality Image Fusion: Unified Framework and 100k Benchmark

Depth and Thermal Sensor Fusion to Enhance 3D Thermographic Reconstruction

3DRIMR: 3D Reconstruction and Imaging via mmWave Radar based on Deep Learning

High-precision Human Body Acquisition Via Multi-View Binocular Stereopsis

RobustFusion: Robust Volumetric Performance Reconstruction under Human-object Interactions from Monocular RGBD Stream

ThermRad: A Multi-modal Dataset for Robust 3D Object Detection under Challenging Conditions

MUC: Mixture of Uncalibrated Cameras for Robust 3D Human Body Reconstruction

An Adaptive Multimodal Fusion 3D Object Detection Algorithm for Unmanned Systems in Adverse Weather

ProbRadarM3F: mmWave Radar based Human Skeletal Pose Estimation with Probability Map Guided Multi-Format Feature Fusion

HeteroFusion: Dense Scene Reconstruction Integrating Multi-Sensors.

RF-based Multi-view Pose Machine for Multi-Person 3D Pose Estimation.