Research map / Point clouds / Segmentation and detection
An LSTM Approach to Temporal 3D Object Detection in LiDAR Point Clouds
Rui Huang, Wanyue Zhang, Abhijit Kundu, Caroline Pantofaru, David A Ross, Thomas Funkhouser, Alireza Fathi
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 the sparse conv LSTM (4 frames) to single-frame U-Net | U-Net detector · Waymo Open Dataset val (vehicles) | mAP@0.7: 56.1 -> 63.6 | Improved |
| use LSTM instead of 4-frame concatenation | multi-frame detector · Waymo Open Dataset val (vehicles) | mAP@0.7: 62.4 -> 63.6 | Improved |
| add Kalman filter tracking on top of single-frame detector | U-Net detector · Waymo Open Dataset val (vehicles) | mAP@0.7: 56.1 -> 56.8 | Improved |
| concatenate 4 frames in the U-Net input | U-Net detector · Waymo Open Dataset val (vehicles) | mAP@0.7: 56.1 -> 62.4 | Improved |
| use the LSTM module on a single frame (ablation) | U-Net detector · Waymo Open Dataset val (vehicles) | mAP@0.7: 56.1 -> 58.7 | Improved |
| increase LSTM input from 1 to 2 frames (ablation) | sparse LSTM · Waymo Open Dataset val (vehicles) | mAP@0.7: 58.7 -> 59.7 | Improved |
| increase LSTM input from 1 to 4 frames (ablation) | sparse LSTM · Waymo Open Dataset val (vehicles) | mAP@0.7: 58.7 -> 63.6 | Improved |
| increase LSTM input from 4 to 7 frames (batch size 1) (ablation) | sparse LSTM · Waymo Open Dataset val (vehicles) | mAP@0.7: 62.3 -> 63.3 | Improved |
| reduce batch size from 2 to 1 for 4-frame LSTM (ablation) | sparse LSTM · Waymo Open Dataset val (vehicles) | mAP@0.7: 63.6 -> 62.3 | Hurt |
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.
- PointPillars — PointPillars: Fast Encoders for Object Detection From Point Clouds (CVPR 2019) · 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
- Fully Sparse 3D Object Detection (NeurIPS 2022 · similarity 0.91)
- Approaching Outside: Scaling Unsupervised 3D Object Detection from 2D Scene (ECCV 2024 · similarity 0.90)
- MGTANet: Encoding Sequential LiDAR Points Using Long Short-Term Motion-Guided Temporal Attention for 3D Object Detection (AAAI 2023 · similarity 0.90)
- LSK3DNet: Towards Effective and Efficient 3D Perception with Large Sparse Kernels (CVPR 2024 · similarity 0.89)
- SAFDNet: A Simple and Effective Network for Fully Sparse 3D Object Detection (CVPR 2024 · similarity 0.89)
- PTT: Point-Trajectory Transformer for Efficient Temporal 3D Object Detection (CVPR 2024 · similarity 0.88)
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.
