InScope: A New Real-world 3D Infrastructure-side Collaborative Perception Dataset for Open Traffic Scenarios

This is the official implementation of the InScope dataset. The paper has been accepted by Information Fusion. "InScope: A New Real-world 3D Infrastructure-side Collaborative Perception Dataset for Open Traffic Scenarios". Xiaofei Zhang, Yining Li, Jinping Wang, Xiangyi Qin, Ying Shen, Zhengping Fan, Xiaojun Tan^†

The ground truth of sequence 0000.

Overview

• Data Download
• Data Loading
• Quick Start
• Benchmark
• TODO
• Citation
• Acknowledgment

Data Download

Due to project restrictions, the InScope dataset is made conditionally public. If you need to use the InScope dataset, please fill in the following ./assets/InScope_Dataset_Release_Agreement.docx file and email your full name and affiliation to the contact person. We ask for your information only to ensure the dataset is used for non-commercial purposes.

After downloading the data, please put the data in the following structure:

├── InScope-Sec, InScope_Pri, and InScope datasets
│   ├── ImageSets
|      |── train.txt
|      |── test.txt
|      |── val.txt
│   ├── labels
|      |── 000000.txt
|      |── 000001.txt
|      |── 000002.txt
|      |── ...
│   ├── points
|      |── 000000.npy
|      |── 000001.npy
|      |── 000002.npy
|      |── ...

├── InScope_track
│   ├── label_02
|      |── 0000.txt
|      |── 0001.txt
|      |── 0002.txt
|      |── ...
│   ├── points
|      |── 0000
|          |── 000000.bin
|          |── 000001.bin
|          |── 000002.bin
|          |── ...
|      |── 0001
|      |── 0002
|      |── ...
│   ├── evaluate_tracking.seqmap
│   ├── evaluate_tracking.seqmap.test
│   ├── evaluate_tracking.seqmap.training
│   ├── evaluate_tracking.seqmap.val

Data Loading

To facilitate researchers' use and understanding, we adapted the InScope dataset to the OpenPCDet framework and provided the corresponding dataset configuration file ./InScope.config

Quick Start

For detection training & inference, you can find instructions in detection_code/openpcdet/README_InScope.md in detail.

All the checkpoints are released in link in the tabels below, you can save them in codes/ckpts/.

Benchmark

Results of 3D object detection based on the InScope dataset

Results of 3D object detection based on the InScope-Sec, InScope_Pri, and InScope datasets

Detection result based on the InScope-Sec Only

Detection result based on the InScope_Pri Only

Detection result based on the Early Fusion (InScope) Mechanism

Detection result based on the Late Fusion Mechanism

Detection result based on the Middle Fusion Mechanism (based on BEV fusion framework)

We provide another middle fusion framework, based on 2D feature fusion mechanism. The checkpoint can be found in [PV-RCNN++ 2D fusion URL] and [CenterPoint 2D fusion URL].

Results of data domain transfer on the car class

3D Multiobject tracking results on the car, pedestrian, cyclist, and truck.

Tracking result of the AD3DMOT based on the InScope dataset on the car class (IoU threshold = 0.5/0.7)

Tracking result of the AD3DMOT based on the InScope-Pri dataset on the car class (IoU threshold = 0.5/0.7)

Tracking result of the AD3DMOT based on the InScope dataset on the pedestrian class (IoU threshold = 0.25/0.5)

Tracking result of the AD3DMOT based on the InScope-Pri dataset on the pedestrian class (IoU threshold = 0.25/0.5)

Tracking result of the AD3DMOT based on the InScope dataset on the cyclist class (IoU threshold = 0.25/0.5)

Tracking result of the AD3DMOT based on the InScope-Pri dataset on the cyclist class (IoU threshold = 0.25/0.5)

Tracking result of the AD3DMOT based on the InScope dataset on the truck class (IoU threshold = 0.5/0.7)

Tracking result of the AD3DMOT based on the InScope-Pri dataset on the truck class (IoU threshold = 0.5/0.7)

TODO

The code and configuration of 3DMOT on the InScope dataset will be released.

Citation

If you find InScope useful in your research or applications, please consider giving us a star 🌟.

The BibTeX format is as follows:

@article{inscope_2026,
title = {InScope: A new real-world 3D infrastructure-side collaborative perception dataset for open traffic scenarios},
journal = {Information Fusion},
volume = {128},
pages = {103951},
year = {2026},
issn = {1566-2535},
doi = {https://doi.org/10.1016/j.inffus.2025.103951},
author = {Xiaofei Zhang and Yining Li and Jinping Wang and Xiangyi Qin and Ying Shen and Zhengping Fan and Xiaojun Tan},
}

Acknowledgment

• DAIR-V2X
• AB3DMOT
• ST3D