Research map / Point clouds / Segmentation and detection
MPPNet: Multi-Frame Feature Intertwining with Proxy Points for 3D Temporal Object Detection
Xuesong Chen, Shaoshuai Shi, Benjin Zhu, Ka Chun Cheung, Hang Xu, Hongsheng Li
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 |
|---|---|---|---|
| add proxy points instead of raw LiDAR points (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 71.13 -> 73.08 | Improved |
| use 3D MLP Mixer instead of self-attention for intra-group mixing (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.71 -> 73.08 | Improved |
| add inter-group cross-attention (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.36 -> 73.08 | Improved |
| add index-based position embedding in cross-attention (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.97 -> 73.08 | Improved |
| use all-group summarization as key instead of concatenating other groups (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.95 -> 73.08 | Improved |
| add motion encoding (decoupled geometry and motion) (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.78 -> 73.08 | Improved |
| decouple geometry and motion encoding instead of integrated time encoding (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.89 -> 73.08 | Improved |
| add boxes' embedding (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.98 -> 73.08 | Improved |
| add trajectory augmentation (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.62 -> 73.08 | Improved |
| add intermediate supervision (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.47 -> 73.08 | Improved |
| use 4x4x4 instead of 3x3x3 proxy points (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 72.54 -> 73.08 | Improved |
| use 5x5x5 instead of 4x4x4 proxy points (ablation) | MPPNet · Waymo val vehicle (3 epochs, 4 frames) | mAPH L2: 73.08 -> 72.98 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PV-RCNN++ — PV-RCNN: Point-Voxel Feature Set Abstraction for 3D Object Detection (CVPR 2020) · better on 5 of 5 shared results
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
- MSF: Motion-Guided Sequential Fusion for Efficient 3D Object Detection From Point Cloud Sequences (CVPR 2023 · similarity 0.89)
- 3D-MAN: 3D Multi-Frame Attention Network for Object Detection (CVPR 2021 · similarity 0.88)
- M3SOT: Multi-Frame, Multi-Field, Multi-Space 3D Single Object Tracking (AAAI 2024 · similarity 0.88)
- Multi-View 3D Object Detection Network for Autonomous Driving (CVPR 2017 · similarity 0.88)
- PTT: Point-Trajectory Transformer for Efficient Temporal 3D Object Detection (CVPR 2024 · similarity 0.88)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.86)
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.
