IMPNet

This is a Pytorch-Lightning implementation of the paper

Efficient Instance Motion-Aware Point Cloud Scene Prediction

Yiming Fang, Xieyuanli Chen, Neng Wang, Kaihong Huang, Hesheng Wang, Huimin Lu

Submitted to IROS 2025.

In this paper, we propose IMPNet, a novel instance motion-aware network for future point cloud scene prediction. Unlike prior works, IMPNet explicitly incorporates motion and instance-level information to enhance PCP accuracy. Specifically, we extract appearance and motion features from range images and residual images using a dual-branch convolutional network and fuse them via a motion attention block. Our framework further integrates a motion head for identifying moving objects and an instance-assisted training strategy to improve instance-wise point cloud predictions. Extensive experiments on multiple datasets demonstrate that our proposed network achieves state-of-the-art (SOTA) performance in PCP with superior predictive accuracy and robust generalization across diverse driving scenarios.

Table of Contents

1. Installation
2. Data
3. Training
4. Testing
5. Download
6. Acknowledgment

Installation

Clone this repository and run

cd impnet
git submodule update --init

to install the Chamfer distance submodule. The Chamfer distance submodule is originally taken from here with some modifications to use it as a submodule. All parameters are stored in config/parameters.yaml.

In our project, all our dependencies are managed by miniconda. Use the following command to create the conda environment:

conda env create -f impnet.yml

Then activate the environment using the command conda activate atppnet

Data

Download the SemanticKITTI data from the official website.

We process the data in advance to speed up training. To prepare the dataset from the our dataset, set the value of GENERATE_FILES to true in config/parameters.yaml.

Training

After following the data preparation tutorial, the model can be trained in the following way:

The training script can be run by

python train.py

using the parameters defined in config/parameters.yaml. Pass the flag --help if you want to see more options like resuming from a checkpoint or initializing the weights from a pre-trained model. A directory will be created in pcf/runs which makes it easier to discriminate between different runs and to avoid overwriting existing logs. The script saves everything like the used config, logs and checkpoints into a path pcf/runs/COMMIT/EXPERIMENT_DATE_TIME consisting of the current git commit ID (this allows you to checkout at the last git commit used for training), the specified experiment ID (pcf by default) and the date and time.

Testing

Test your model by running

python test.py -m COMMIT/EXPERIMENT_DATE_TIME

Download

Please download the model file from (here)

Acknowledgment

We would like to express our sincere gratitude for the following open-source work that has been immensely helpful in the development of IMPNet.

• ATPPnet: ATPPNet: Attention based Temporal Point cloud Prediction Network.
• PCPNet: PCPNet: An Efficient and Semantic-Enhanced Transformer Network for Point Cloud Prediction.

License

This project is free software made available under the MIT License. For details see the LICENSE file.