The problem Interstice solves

The crime rates are increasing exponentially. India is toppping the charts for wrong reasons.
Several CCTV cameras and surveillance systems are being installed all across the country, but it is still proving to be inefficient.
For example in Mumbai, a total of 5000 Govt. operated CCTV cameras were installed over 1510 locations.
Yet in the year 2020, 148 murders , 4539 criminal cases against women , 938 against children were registered in the city with CCTV security.
Is there a way to minimise this?
What if the crime/violence is reported in realtime using technology.
In this project we try to explore this possibility. Introducing to you Project Interstice.

Our model uses Computer Vision to narrow down to the target for faster and more reliable predictions. We used OpenCV, a Computer Vision library, to track the movement of people in real time using Multiple Object Detection through Localisation. Deep Learning models like Customised CNN Model, VGG Net, AlexNet, Inception+CustomisedCNN were trained on a dataset which consisted of around 2200 videos. The dataset we used was seeded from combined datasets scrapped from different sources, comprising videos taken from various angles, enclosing all types of violent and non-violent natured clips.

Challenges we ran into

During this project, we ran into several problems. Most prominent of them were:

1. Finding the Dataset: Violence and Non-violence being a very subjective matter, it was very hard to find a suitable dataset for training the model.
2. Mode of Presentation: Being a single script, we had to decide the mode of presentation. Web Platfrom being the most prominent and wide-spread platform, we chose it.
3. Hosting: Hosting the tensorflow model on a website had its own issues.