Automated Censorable Content identification in Films Using Deeplearning

Akhil G Krishnan — Wed, 20 May 2020 16:40:23 +0000

What is my project ?

The movies may contains censorable scenes like smoking, heavy drinking, riding without a helmet, driving without the seatbelt, etc. It is the responsibility of the editors to check the whole video to identify such contents. This process consumes a large amount of time. We introduce an automated statutory warning generator that will add exact statutory labels to the scene. This is more helpful for movie editors.

Demo Link

https://drive.google.com/open?id=1sf1FDhFkoBceAA7SenRlr8AomZ_f_8l4

Link to Code

AkhilGKrishnan / CensorLabel

Automated censorable content identification system for movie using deeplearning

CensorLabel

Automated censorable content identification system for movie using deeplearning. This project automatically add statutory warnings in movies when any censorable contents where shown in the movies. The Project is developed using YOLOV3 and OpenCV...

This system is used for smoking detection, Check whether the motorcyclist weared helmet or not, driving without wearing seatbelt and alcohol detection.

View on GitHub

How I built it

In this project we used two deeplearning model, first is ResNet model for Human Activity Recognition and the second is YoloV3 model for object detection. And we used openCV dnn module for detection. We created an desktop application for users to add videos. The desktop application is built using python EEL.

Result

We got a better result in accurac. We got 85% of mAP , so we can say that this system will perfect detect the censorable contents in movies and add corresponding statutory warning to that scenes.

Thanks and everyone stay healthy and safe.

DEV Community: Akhil G Krishnan