CUNY Hackton project description:
Idea: Visually impaired people still have inconvenience when they travel around because there aren’t enough technologies to support those people. Computer vision recognition might be a solution for supporting visually impaired people. Computer vision recognition nowadays is available by algorithms of machine learning and deep learning. Cameras and sensors become extremely useful and powerful to help computers and mobile phones to do image recognition. Companies like Google, IBM, and Apple already develop a bunch of Application Programming Interface(API) to recognize objects like facial expression, foods, table, etc. Our idea is by using the Google cloud vision API to help us build a demo that enables visually impaired people to detect obstacles when they travel around.
Progress: We first discussed and determined our idea and goal that doing something good for our society. We planned to make a demo that can recognize some common objects from images and gives feedback to help visually impaired people. Then, We research google cloud computer vision API to check if we can use it to build our project. We select some images online that contain objects such as traffic lights, stairs, and cars that may be important for visually impaired people to be aware of. By invoking google cloud computer vision API, we are able to detect objects and label some details. In order to give the user audio feedback, we then search for a voice package that can use on python and put the voice library into our script. Moreover, we provide some features that highlights the common object and determines whether the object is on the left or right.
Challenges: Learning and invoking google cloud API at a limit of time filtering some unnecessary information Seeking details for our projects Integrating different ideas Time management Problem-solving
Future work: Wearable smart devices such as google glass and other smart watches are pretty desirable technologies for helping visually impaired people. Hopefully, we can make our project to detect real-time surrounding objects and eventually put our works on smart glasses. Aslo, we can train a model, which corrects all necessary data, like some obstacles probably make danger for them. Our goal is to provide a more safe and convenient way to help visually impaired people when they travel around.
Log in or sign up for Devpost to join the conversation.