Imagine this: you're at the Ubahn station and after several minutes of waiting, finally your train arrives. However, nearly all the cars are filled with people and you just can't get in! This can/has possibly been a common occurrence for any of us during peak hours of the day.
What it does
Our solution is to give you, a passenger waiting at the station, an overview of the capacity of the cars of the incoming train. This way you see which cars are too full to enter and which ones are less filled and easier accessible. From this, you decide what part of the platform to strategically position yourself in order to access a less crowded car or to wait for the next train
How we built it
Our application was built using OpenVino, OpenCV, Python and on top of base code provided by Intel
Challenges we ran into
- Inaccessibility to ready data hence we had to create these ourselves by going into the Ubahn to film videos for image detection
- The fluctuation in the number of segmentation per frame
- Image detection didn't not work for people obscured by a glass panel
- Drawings on posters in the trains were detected as well
- Installation limitations from OS
Accomplishments that we're proud of
- Successfully running the image detection model ourselves
- Successfully processing videos to find the free places in each car of the train.
What we learned
- Thought process required to brainstorm to identify a vital problem and to come up with a viable solution
- Image processing using openCv and python.
- Creativity towards data collection process
What's next for Let me in!
- Further develop the image detection model to exclude unwanted objects like posters and to detect people behind glass
- Further development for usage in crowd control during events that attract lots of people