False Eyes
Imagine for a minute what life would be like if your vision was permanently impaired and could not be corrected with eyeglasses, contact lenses or vision surgery. This is the situation many people find themselves in every day, suffering from what’s called “blindness” or ‘low vision”. I see many such persons at metro stations or bus stand who are struggling to find a way to board on the metro or bus. Sometimes they fell from the bus or metro during boarding and get hurt. It feels bad for them. We can’t get back their vision but we can make something which will help them to live a little better.
My idea is for blind persons. As they don’t see with their eyes, Can’t we replace it with False Eyes? Which means can’t we replace with cameras which are high tech these days? They don’t have eyes but their other organs are extremely active such as ears, brain, visualization, etc.
What we can do is to convert images to speech so that the blind person can hear it and visualize where the things/objects oriented or moving. If a blind person is walking then it will notify about the objects which occur in their path or find a way to make a track. My system can guide him/her to find a way so that they can reach their destination.
It can be done using Convolution Neural Network and Recurrent Neural Network.
Here, we are making spectacles like object where two cameras will be placed at the place of the lens. Two cameras because it will be easy for majoring the distance from a center point or person to object. Each camera capturing the images and encoding the objects which can translate through ears phones which are connected to spectacles. It can also measure distances or movement of objects in a 3D plane. Suppose an object is approaching to person then, the image formed in the spectacle appear bigger with some ratio. We can also find this using vectors, in which direction they are going and what speed they must have. So that they can find their own way without collision. And they can also find the location such as a bus stand or their home or where they work. We can train our model for different places such as their homes, office, etc. It will help them a lot.
It has one disadvantage, our system should be perfect 100% if somewhere our system fails it can lead to a major problem so, have to take care of it.
#machinelearning #deeplearning #dnncompiler #artificialintelligence
