PDIWS: Thermal Imaging Dataset for Person Detection in Intrusion Warning Systems

Citation Author(s):
Nguyen Duc
Thuan
Hanoi University of Science and Technology
Le Hai
Anh
Hanoi University of Science and Technology
Hoang Si
Hong
Hanoi University of Science and Technology
Submitted by:
Nguyen Thuan
Last updated:
Fri, 08/11/2023 - 05:25
DOI:
10.21227/v02g-rr56
Research Article Link:
License:
0
0 ratings - Please login to submit your rating.

Abstract 

This dataset presents a synthetic thermal imaging dataset for Person Detection in Intrusion Warning Systems (PDIWS). The dataset consists of a training set with 2000 images and a test set with 500 images. Each image is synthesized by compounding a subject (intruder) with a background using the modified Poisson image editing method. There are 50 different backgrounds and nearly 1000 subjects divided into five classes according to five human poses: creeping, crawling, stooping, climbing and other. The presence of the intruder will be confirmed if the first four poses are detected. Advanced object detection algorithms have been implemented with this dataset and give relatively satisfactory results, with the highest mAP values of 95.5% and 90.9% for IoU of 0.5 and 0.75 respectively. The dataset is freely published online for research purposes at https://github.com/thuan-researcher/Intruder-Thermal-Dataset.

 

Instructions: 

Each image in the dataset is compounded by an object image and a background using the Poison image editing method. The dataset folder ./PDIWS consists of two subsets train and test:

  • train: 2,000 images under .JPG format, each image contains only one object.
  • test: 500 images under .JPG format, each image contains only one object.

Labels for those images are stored in files train.json and test.json. Each contains a list of annotation directories following the format:

{"image_id": i, "bbox": [x, y, w, h], "class": c}

 

where i is the image index (file name), bbox is the bounding box, class (0-4) is the type of position (0: creeping, 1: crawling, 2: stooping, 3: climbing, 4: other).

 

Funding Agency: 
Vingroup Innovation Foundation (VINIF)
Grant Number: 
VINIF.2022.ThS.086
Data Descriptor Article DOI: 

Comments

For more information, please contact me via email: thuan.nguyenduc1@hust.edu.vn.

Submitted by Nguyen Thuan on Thu, 01/25/2024 - 22:52

666

Submitted by Guancheng Zhu on Tue, 03/12/2024 - 06:55

For more information, please contact me via email: thuan.nguyenduc1@hust.edu.vn.

Submitted by Nguyen Thuan on Thu, 01/25/2024 - 23:14