The 3BlindMice
Our project helps blind people in society. We created an app that detects obstacles and says what it is. The app uses a ResNet-50 as a training model to detect the objects that are in front of the user. This app finds the input that is gained from the training model and then says it out loud for the user to know what is in front of it. It will only say it out loud if the confidence level of the training model is over 45%. It will also rumble when the threshold is met. We took the help of Brian Voong's code on real-time object detection from youtube. However, we modified the code to let it help the blind by limiting the threshold for the accuracy of the objects detected, converting text to speech and correcting the colors for easy readability for the partially blind. Unfortunately, the dataset (ResNet 50) is not powerful enough for more accurate object detection. We are investing further time to improve our app's accuracy by researching on better data sets.
Log in or sign up for Devpost to join the conversation.