search

yichen928 / SparseFusion

[ICCV 2023] SparseFusion: Fusing Multi-Modal Sparse Representations for Multi-Sensor 3D Object Detection
Apache License 2.0
198 stars 19 forks source link

Semantic information transfer #12

Closed zhaowenZhou closed 1 year ago

zhaowenZhou commented 1 year ago

I'm wondering in which step was the semantic information fusion done? It seems like it was done in HardVFE but in your README file, the model is sparsefusion_nusc_voxel_LC_r50.py which used HARDSimpleVFE, it doesnt contain semantic information fusion

yichen928 commented 1 year ago

Hi. Thanks for your interest. The semantic information transfer is done in generate_heatmap_deform of file mmdet3d/models/dense_heads/sparsefusion_head_deform.py https://github.com/yichen928/SparseFusion/blob/89a7c5a9e2a1bd1bf63631819af8a78c69df744c/mmdet3d/models/dense_heads/sparsefusion_head_deform.py#L598