Buy Now ₹1501

ML powered system for detecting weapons within images

Business Problem

1. Mass shootings have become increasingly prevalent at public gatherings

• Creating an algorithm that that be integrated into traditional surveillance systems can be used to detect threats faster and more efficiently than those monitored by people
• In modern surveillance systems, there is a person or group of people, in charge of watching monitors which can span across multiple floors of a given area

2. Violence on social media platforms such as Youtube, Facebook, and TikTok

• An algorithm that integrate itself into traditional upload systems can detect violent videos before they are spread on a given website
• Considering the graphs below, the United States ranks among the top 5 countries in terms of firearm deaths
  • Total Deaths Data
  • Mass Shootings Data

Solution

1. Create a neural network that can be integrated into traditional surveillance systems
2. This neural network will be able to detect whether a firearm is present in a frame, and if so, it will notify authorities/managers of its detection

Requirements

1. keras (PlaidML backend --> GPU: RX 580 8GB)
2. numpy
3. pandas
4. opencv (opencv-contrib-python)
5. matplotlib
6. beautifulsoup

Datasets

• Data used in this project can be found on my Google Drive
  • Weapons
  • No Weapons