Research map / Detection and segmentation / 3D detection and driving
HENet: Hybrid Encoding for End-to-end Multi-task 3D Perception from Multi-view Cameras
Zhongyu Xia, ZhiWei Lin, Xinhao Wang, Yongtao Wang, Yun Xing, Shengxiang Qi, Nan Dong, Ming-Hsuan Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use hybrid image encoding (large short-term + small long-term encoder) instead of increasing resolution (ablation) | HENet · nuScenes val 3D detection | NDS: 59.2 -> 59.9 | Improved |
| ensemble two baselines by NMS instead of hybrid encoding (ablation) | BEVDepth4D + BEVStereo · nuScenes val 3D detection | NDS: 57.3 vs 59.9 (HENet) | Hurt |
| add AFFM temporal integration (ablation) | HENet · nuScenes val 3D detection | NDS: 59.2 -> 59.9 | Improved |
| add AFFM temporal integration (ablation) | HENet · nuScenes val multi-task | BEV seg mIoU: 56.9 -> 58.0 | Improved |
| use adjacent frame attention instead of global attention (ablation) | BEVDepth4D temporal fusion · nuScenes val | NDS: 52.6 -> 53.1 | Improved |
| use attention-based AFFM instead of concatenation and convolution (ablation) | BEVDepth4D temporal fusion · nuScenes val | NDS: 52.8 -> 53.2 | Improved |
| add backward pass to forward AFFM (ablation) | BEVDepth4D temporal fusion · nuScenes val | mAP: 41.2 -> 41.5 | Improved |
| use a larger BEV encoder in place of AFFM (ablation) | BEVDepth4D temporal fusion · nuScenes val | NDS: 52.3 -> 52.4 | No clear effect |
| use a larger BEV grid (0.8m) for both tasks (ablation) | multi-task BEVDepth4D · nuScenes val | NDS / mIoU: NDS 53.2 -> 50.5; mIoU 41.6 -> 50.9 | Mixed |
| use 0.4m grid for detection and 0.8m grid for segmentation (ablation) | multi-task BEVDepth4D · nuScenes val | mIoU: 41.6 -> 50.6 | Improved |
| add adaptive feature selection (ablation) | multi-task BEVDepth4D · nuScenes val | NDS: 52.9 -> 53.3 | Improved |
| add independent BEV encoder per task (ablation) | multi-task BEVDepth4D · nuScenes val | mIoU: 51.2 -> 54.0 | Improved |
Later papers that compared with this method
1 later accepted paper reports HENet as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Breaking Alignment Barriers: TPS-Driven Semantic Correlation Learning for Alignment-Free RGB-T Salient Object Detection (AAAI 2026 · better on 6 of 6 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 4 on most shared results.
- SparseBEV — SparseBEV: High-Performance Sparse 3D Object Detection from Multi-Camera Videos (ICCV 2023) · better on 2 of 2 shared results
- Far3D — Far3D: Expanding the Horizon for Surround-View 3D Object Detection (AAAI 2024) · better on 1 of 1 shared result
- PETRv2 — PETRv2: A Unified Framework for 3D Perception from Multi-Camera Images (ICCV 2023) · better on 0 of 1 shared result
- BEVFormer — BEVFormer: Learning Bird’s-Eye-View Representation from Multi-Camera Images via Spatiotemporal Transformers (ECCV 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- MatrixVT: Efficient Multi-Camera to BEV Transformation for 3D Perception (ICCV 2023 · similarity 0.87)
- Fast-BEV: A Fast and Strong Bird’s-Eye View Perception Baseline (TPAMI 2024 · similarity 0.87)
- M-BEV: Masked BEV Perception for Robust Autonomous Driving (AAAI 2024 · similarity 0.86)
- M3Net: Multimodal Multi-task Learning for 3D Detection, Segmentation, and Occupancy Prediction in Autonomous Driving (AAAI 2025 · similarity 0.86)
- SOGDet: Semantic-Occupancy Guided Multi-View 3D Object Detection (AAAI 2024 · similarity 0.86)
- The 3D-PC: a benchmark for visual perspective taking in humans and machines (ICLR 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
