Crime Video Dataset for Indian Scenarios

Citation Author(s):
Saylee
Begampure
MIT World Peace University
Parul
Jadhav
MIT World Peace University
Submitted by:
Saylee Begampure
Last updated:
Tue, 01/25/2022 - 05:16
DOI:
10.21227/e2yd-k649
Data Format:
Links:
License:
5
1 rating - Please login to submit your rating.

Abstract 

 The crime rate is increasing at a high rate in India. Terrorist attacks like Mumbai 26/11, Pulwama attack, Pune German Beckary attacks have created terrific fear amongst Indian Society. Video analytics plays a significant role in detecting and predicting such suspicious human activities using deep learning models It will help in reducing the increasing crime rate by preventing treacherous actions. Video analytics analyzes the video content and adds brains to eyes i.e. analytics to the camera. It extracts contents from the video by monitoring the video in real-time. There is a huge advancement in human action recognitions and prediction-related research but still, state of art algorithms needs improvement in classifying it. The main objective here is to create a dataset for Indian scenarios for different crime actions like Fighting, Abuse, Arrest, Arson  & Burglary. It can then be used to train a deep learning model to detect and predict criminal actions reducing terrorism in India.

Instructions: 

============================= A few notes on using the dataset ===========================

Saylee Begampure ,Parul Jadhav

https://mitwpu.edu.in/

sayalee.b17@gmail.com

parul.jadhav@mitwpu.edu.in

 

Cite our paper:

Saylee Begampure, Parul Jadav, “Intelligent video analytics for human action detection: a deep learning approach with transfer learning"

Elsevier Scopus indexed, International Journal of Computing and Digital Systems", University of Bahrain, 2021-07-14

=====================================================================================

Dataset consists of following files:

 

1. Crime_sites.docx

It contains sources of crime sites from where Indian scenario videos are taken.

Categories considered here are

1.Abuse 2. Arrest 3. Fighting 4. Burglary & 5. Arson

Videos are searched on YouTube from the above category.

Most of the videos considered are from CCTV footages, News channels, and movies and

as it is created for Indian scenarios videos from various states from India like Punjab, Maharashtra is considered.

 

2. Original Videos

It contains entire footage downloaded from the above YouTube links.

Videos are saved in .mp4 format.

It has crime actions of different categories from different parts of India like Panjab, Delhi, etc.

Downloaded Videos are saved into respective Folders.

 

 

3. Cropped Videos

The downloaded video from YouTube has a lot of information.

The part of the video where crime is happening and which can make or break the case lies in a small part of it,

so it was important to crop the particular portion of the video.

The Online Video cutter platform is used to crop the part of the video where the crime was present.

Cropped Videos of the smaller and required sizes are saved into their respective folder.

 

4. Annotation of the videos

Lastly, all cropped videos are annotated as per the crime category.

Naming Convention: Name of the category _ Number of Video

e.g for 1 st video from the Fighting Category: Fighting_001

 

=====================================================================================

Comments

Indian Crime Videos Dataset

Submitted by Saylee Begampure on Wed, 11/10/2021 - 08:09