Technology can be of assistance in incredible ways. Please, check the illustrations in the video. There are several apps for people with visual impairment. They can read text, recognize objects and even count money (cash). However, this is not enough. What about letting a blind person receive a meaningful description of the scene in front of them? What if we return them the possibility to see through the ears. Current state-of-the-art methods could make this happen. In this project we just try to make it real.

What it does

By using a wearable camera, visual information is captured and processed in order to generate a description of what appears in the image. This description is told to the blind person so he/she could know what is happening in front of them. Of course, it is not perfect and has mistakes, but

How I built it

Using existing research and implementing it on tensorflow. It required to train for long and of course a lot of data!

Challenges I ran into

Training time is long for the timing of a Hackathon. Training further would provide better results.

Battery usage and memory constraints due to limitations of wearable devices

Accomplishments that I'm proud of

Use technology to try to help people in need

What I learned

Technology already exists: let's use it! Current research is amazing and advances quickly. Let's bring it to reality! Research is under constant improvement and there are mistakes, but for sure the current state is better than not having anything like that at all!

What's next for I can see

Bring this to people by getting a funding!
Keep improving and keep updating this assistance following updated advances in science.

Built With

  • gpu
  • python
  • smartglasses
  • tensorflow
+ 47 more
Share this project: