The problem Surakshit solves

2 of the major problems in the crime sector is providing security and assurance to the victims who are forced to take down crime records when a powerful or influential person is involved on the other side, and if they don't agree to it, they even go to an extent of destroying those records.
The second is measures to prevent crime. Though there are a lot of CCTV cameras deployed across the country, they are majorly used to identify evidences after a crime has occurred, rather than using them to prevent the crime in the first place. Also, the lack of sufficient police forces leads to all regions not being secure.
We'd like to implement an application which uses technology to address all these issues.

Blockchain Smart Contracts can be used to decentralize and secure crime records, such that victims may not be pressurized to take the cases down, as it frequently happens in sensitive cases such as rape. Statistics show that crime usually happens in regions with very less activity, therefore by developing a model which can detect activity in a regions(for example, count the number of vehicles and people) and also regions with extreme and emergency situations such as fighting, mob rage etc., and send notifications about regions with such activity, police patrolling can be done in those regions. Hence this will not only be beneficial in preventing crimes, but also effectively allocate police forces on their patrolling duty.

Challenges we ran into

Finding a large video data set to train the model and the lower computer specifications during training period and also the model was a bit slow because of our system specifications, therefore we had to compress the videos before feeding it to the model.
Also, this was our first time experimenting with the Matic Network, so we ran into quite a few errors while deploying the Dapp on the Matic Network, but after thoroughly going through the documentation, we were able to overcome this issue.
This was also the first time we dealt with Web2py to integrate our models with the client side, so we ran into many errors in that as well. Using our experience in Web Development, and with the help from stackoverflow, we were able to solve these issues as well.