Awesome

[Towards]

This repository is the official HiXray dataset.

Download link

(China mainland, BaiduNetdisk) https://pan.baidu.com/s/1ugMhHtEp4G7yEF_WbLt8DA password: buaa

(Other area, Google Drive) https://drive.google.com/drive/folders/1zlDLw36SQp5Y9Oi-TzUmZc4zFKk6nHgW?usp=sharing

HiXray dataset:

LIM framework:

Table of Contents

• Dataset
• Install
• Results
• Acknowledgement

Dataset

In this project, we build the largest High-quality X-ray Security Inspection dataset.

• All images of HiXray dataset are annotated manually by professional inspectors from an international airport, and the standard of annotating is based on the standard of training security inspectors.

• HiXray dataset contains a total of 45364 X-ray images(36295 for training, 9069 for testing), including 8 categories of cutters, namely, 'Portable_Charger_1','Portable_Charger_2','Mobile_Phone','Laptop','Tablet','Cosmetic','Water','Nonmetallic_Lighter'

  <div align=center> <img src="images/statistics.png"/> </div>

• The information structure of annotation file is as follows: image name, category, top-left position of prohibited item (x1, y1), bottom-right position of prohibited item (x2, y2).

• Comparison of existing open-source X-ray datasets:

  <div align=center> <img src="images/dataset_comparison.png"/> </div>

Install

1. If you want to train your model, execute the following command: change root to MuBo

   cd MuBo
   

   Change the value of HiXray_ROOT variable in MuBo/data/HiXray.py file to the path where the training set is located, for example,

   HiXray_ROOT = "/mnt/cvpr_dataset/train/"
   

   run train.py

   python train.py --save_folder /mnt/model/Mubo/save/ --image_sets /mnt/cvpr_dataset/train/train_name.txt --transfer /mnt/ssd300_mAP_77.43_v2.pth
   

   save_folder is used to save the weight file obtained by training the model,

   image_sets is the path to a TXT file that saves all the picture names used for training,

   transfer indicates the pre-trained weight of SSD on VOC0712 (available at here).

2. If you want to test our model, execute the following command: change root to MuBo

   cd MuBo
   

   Change the value of HiXray_ROOT variable in MuBo/data/HiXray.py file to the path where the testing set is located, for example,

   HiXray_ROOT = "/mnt/cvpr_dataset/test/"
   

   run test.py

   python test.py --trained_model /mnt/model/Mubo/weights/Mubo.pth --imagesetfile /mnt/cvpr_dataset/test/test_name.txt
   

   trained_model is the weight file you want to test, our model is available at here, password is ectq

Results

Comparing with Feature Pyramid Mechanisms:

Visualization of the performance of both the baseline SSD and the MuBo-integrated model:

Citation

If this work helps your research, please cite the following paper.

@inproceedings{Tao:ICCV21,
  author    = {Renshuai Tao and Yanlu Wei and Xiangjian Jiang and Hainan Li and Haotong Qin and Jiakai Wang and Yuqing Ma and Libo Zhang and Xianglong Liu*},
  title     = {Towards Real-world X-ray Security Inspection: A High-quality Benchmark and Lateral Inhibition Module for Prohibited Items Detection},
  booktitle = {IEEE ICCV},
  year      = {2021},
  }