Perceptron_3038

Create-A-Thon 2020 Project Repository

---

Workflow Pipeline :

# Create-a-thon 2020 Final Project : Team : Perceptron_3038

Project title : Facial Recognition in Criminal Detection.

About Sapio Analytics Sapio is a team spread across India, UK, and Silicon Valley, focused on creating high end data based algorithms powered by AI, on the path to become the first purely artificially intelligent policy making system in India and probably the world. Its on its path to become the go-to product for any government body to take decisions- the go-to product that all stakeholders (government representatives, citizens, data experts) would want the government to use before taking decisions.

Problem Statement: Build a face recognition analytics system for law enforcement agencies

Problem Description:

The security of India is becoming ever more critical in this new world. There is an expectation of a slew of organized crime activities that may attempt to cripple our law enforcement and defense agencies. In such a scenario, technology becomes of paramount importance.

One aspect of the use of technology in law enforcement is video analytics, and video analytics has various components. One of the components is to identify patterns across different videos.

The task here is to analyze a 30 second- 1 min video, and capture faces of the people in the same, and then analyze another video to see how many of the persons in the first video are seen again in the second video and at what times.

Now, if the above technology is used for thousands of videos gathered from hundreds of intelligence sources, with cross-checks across all videos, really significant information on patterns in any form of organized crime can be identified.

Allowed Tech Stack Use:

Python OpenCV (even if you are not good in Python, we recommend you try in whatever you can)

Steps in the challenge:

Get yourself a token from Registration deck Attend the inaugration Read and Understand the problem Statement Start developing Attend Shout-outs Check-in code at stipulated checkpoints Finish your code Participants will be shortlisted on 5th July Shortlisted participants will get a chance to work on maturing their solution for the next 5 days After 5 days, finalists will be announced

Challenge format:

Phase 1: 12-hour hackathon Shortlisted ones will get a chance to participate in Phase 2 Phase 2: 5-day hackathon

Eligibility:

Your team should be less than 5 members Your team should have at least 1 Internity intern

FAQs

Should we create a table for getting the details of the candidate You can use anything you want(RDBMS or NoSQL), what matters is an information storage and retrieval What will be in the next phase In the next phase, integration with other platforms like Github, Kaggle, Hackerrank and other competitive programming platforms and make the algorithms more mature to get the best results Who should I connect for assistance Please connect with your coaches or the Create-A-Thon execution team.

Task 1 Analyze a 30 second- 1 min video, and capture faces of the people in the same.

Task 2 Analyze another video to see how many of the persons in the first video are seen again in the second video and at what times.

Task 3 if the above technology is used for thousands of videos gathered from hundreds of intelligence sources, with cross-checks across all videos, really significant information on patterns in any form of organized crime can be identified.

Acceptance Criteria :

Technology/ Models used
Performance of model
Model selection
Scalability
Scalability of model
Retrainable model
Accessibility
User friendly
Accuracy

Demo Snapshots