When police are looking for a criminal through monitors and cameras records. The information is far separate from each other. A criminal could appear in distinct camera in the building or city. This inspired us to create an efficient application to trace and predict the future direction of criminal.
What it does
The project allowed the user to upload video files, decide numbers of cameras, precision level. Then it analyses if there are elements required by the user appeared in the video file. A default map of the building has been constructed using React. The user can design a customized two-dimension map for every place where the camera is located. Our App will trace the route of the element, which is targeted by our user, and the time when the camera captures the element will be presented in a table for further investment.
How I built it
Challenges I ran into
There are a lot of challenges we had encountered . The most difficult challenge for us is to detect the required element from video. We start with detecting the requird elements for a picture, and applied the methodology to video by diving video into frames. In the end, we are able to detect whether the person appeared in the video,
Accomplishments that I'm proud of
What I learned
What's next for Path_Tracing
There are a few things we can improve in the future:
- Single target: We can have another mode to detect two type of object or multiple types of object as long as the API supports.
- Specified target: in the future if we have enough resources. We can apply the customized training for user's own model to find a specific target. This can be achieved by training the API customized module. For e.g: we can use this to find a lost person/kid by access every camera in the city. And used the camera record to trace the route of the person. This relies on a large number of picture of the targeted object.
- Smart distinguish: we will combine the result of different modules to distinguish the target details. This can highly improve the actual precision. If the user inputs a complicated item which is not existing in API’s database. Our app will use the description of the item like color, shape, pattern to search close items.
- we want our app not only indicates the route of the targeted object under the camera, but it should also be able to give a possible/shortest route of the target when it is moving in an unmonitored area. We can do this by embed google map or the detailed building Blueprint in our application.