Technology is helping our lives from improving communication to working more efficiently. However, we already have some negative feelings when we are waiting too long at bus stops. Our ambition is making bus stops more friendly, to make it human-centric service.

What it does

  • The system includes a web camera, a big display, and a web server. The camera on the display can capture the faces of passengers at bus stops and the system can recognize their emotions. Based on detected emotions, the system can optimize advertisements on the display.

  • For example, when a passenger is waiting for a bus on a dark and cold night, he/she would have negative emotions such as anger, fear and loneliness. At that time, the system can detect his/her negative emotions and display a fire place that will relax him/her. Once he/she has positive emotions, the following advertisement can be much more effective for branding. Moreover, the data of passengers' emotions can improve the quality of advertisements.

How we built it

  • Our framework includes three main components: a webcam, a local website, and the Google Cloud Vision API. Every second one image is captured from the webcam and is sent to Google Cloud Vision API using Python. This API returns the number of faces and their emotions to the bus stop (a website developed with Javascript and vue.js).

  • These emotions include: joy, sorrow, surprise and anger. The advertisement will be showed based on how majority of the passengers feel. Furthermore, the number of people and their feeling are showed in the monitor through emoji icons. The system automatically shuts down after 10 second if no one is at the bus stop to save energy.

  • Finally, all data about persons emotions will be recorded to further purpose, for example to analyse how people react to advertisements.

Challenges we ran into

  • First, it cost us three hours to use the GoPro camera for capturing images. The main problems we had was connecting via wifi, controlling the device and saving the high quality image. However, we gave up on this device because we can only connect to one wifi network at the same time and we also need to use Internet connection to call the cloud servicess.

  • Secondly, The quality of the webcam imageswas not good enough (low resolution and small image size). Also, the webcam is placed at top of the bus stop and it's not an ideal viewpoint for face recognition system. Our initial API from Microsoft Cognitive Services easily failed and didn't detect any faces.

  • Thirdly, we got really hungry in the middle night and could not focus to do more coding.

Accomplishments that we're proud of

We’re proud of the broad effects of this system. The system has the merits for three stakeholders, including passengers, JCDecaux, and Advertisers (clients of JCDecaux). For passengers, the system can make them feel relaxed, thanks to optimised contents. For JCDecaux, they can maximise the effect of advertisements and get valuable user data. For Advertisers, the system can contribute to establish strong brands.

What we learned

We learned the importance of teamwork. At first, we had difficulty of communicating with each other because all of us have different backgrounds. After we identified the purpose of this system and divided our roles, it became much easier to communicate.

What's next for HappyBusStop

In the near future, bus stops would become half public and half private space, thanks to the personal mobility technology. The next step of HappyBusStop might be how to create the personalised spaces.

Built With

+ 1 more
Share this project: