Dead - 64 people
Dead - 37 people
Dead - 14 people
Large number of deaths have resulted from fires at public outlets like malls, restaurants and departmental stores in recent times. For safety, especially those in retail, understanding foot traffic is vital. It can also serve as a factor for allocating resources better in case of emergencies. A place with more people require improved rescue services than a place with less people.
What it does
We built a system of web cameras for counting people entering a given area. For eg, a mall or a restaurant is usually enclosed around finite area with multiple entrance and exit points. We add a web camera at each entrance and exit point to maintain a foot count of people in that area. This can help -:
- Allocate the optimal number of rescue personnel we need to deal with an immediate emergency.
- Provide enough data for rescue teams to take informed decisions in the long run.
- Customize evacuation procedures based on variation in footfall with time.
- Know a near exact count of people waiting to be rescued.
How we built it
Entrance/Exit Points: We monitor movements at all entrance and exit points surrounding a given area. This is done using a security camera.
People Detection: We use openCV deep neural network along with a Mobile SSD (Single Shot Multi Box Detector) pre-trained caffe model to detect people. The model has already been trained and run on multiple objects so as to be able detect people. It provides us with a confidence value and classification value to which the confidence adheres.
Challenges we ran into
Accomplishments that we're proud of
What we learned
What's next for CountMeIn