The problem Crime Decoder solves

To enhance and generate a clear image and video to help ease the detection and verification of a crime scene and store the details in a structured way. Although there are many software available, we have handpicked the best from the lot for the same, further, there is absolutely no software which houses both image enhancement and video restoration, as video restoration takes a lot of time, our system is comparatively faster as it uses images instead of videos and gives the user the benefit of applying various models at a particular snap, making detailed analysis possible, where traditional software takes 20-30 minutes, our design takes a max of 10 mins to achieve the same, this could be brought down considerably by improvement in hardware,as here we are using an API because of hardware limitations and shortage of time. As the target audience is the police so we have designed a sleek interface as well as a handy user guide, which would also be available in the regional language(Assamese) as well.

Link to a ppt explaining all the features, models used, datasets used :

https://www.canva.com/design/DAFXG1mVlAE/P0X_VSjCmUuHCyMm0IQn4w/view?utm_content=DAFXG1mVlAE&utm_campaign=designshare&utm_medium=link2&utm_source=sharebutton

Challenges we ran into

1.We tried implementing this on our machine on a smaller dataset but we were not able to use them due to lack of required gpu and also as we used a small portion of the dataset our accuracy wasn’t upto the mark,hence we decided to go with an API.
2.Integrating the various image processing models we used
3.Splitting the video into frames and then recombining them, for this we had to use an open source tool called FFMPEG.