Weapons Detection in Visual Data

To overcome these problems, we propose Orientation Aware Weapons Detection algorithm which provides oriented bounding box and improved detection performance of weapons. The proposed model provides orientation not only using angle as classification problem by dividing angle into eight classes but also angle as regression problem. For training our model of weapon detection, a dataset comprising of more than six thousand weapons images is gathered from the web and then manually annotated with position oriented bounding boxes. The proposed model is evaluated on this dataset, and the comparative analysis with off-the shelf object detectors yields superior performance of the proposed model.

The dataset and the implementation is publicly available at at this github link.

Methdology

Faculty

Dr Muhammad Moazam Fraz
Dr Muhammad Moazam Fraz
Dr Muhammad Shahzad
Dr Muhammad Shahzad

Students

Nazeef ul Haq
Taufial Sajjad Shah

Selected Publications

NU Haq, TS Hashmi; M. M. Fraz; M. Shahzad , “Orientation Aware Weapons Detection In Visual Data: A Benchmark Dataset”, in Computing Volume March, 2022
NU Haq, TS Hashmi; M. M. Fraz; M. Shahzad , “Rotation Aware Object Detection Model with Applications to Weapons Spotting in Surveillance Videos”, Proceedings of the 2021 International Conference on Digital Futures and Transformative Technologies (ICoDT2), May, 2021, Islamabad , Pakistan.
TS Hashmi, NU Haq, M. M. Fraz, M. Shahzad , “Application of Deep Learning for Weapons Detection in Surveillance Videos”, Proceedings of the 2021 International Conference on Digital Futures and Transformative Technologies (ICoDT2), May, 2021, Islamabad , Pakistan.