Around the world, 285 million people are estimated to be visually impaired, with 39 million of them being blind (World Health Organization). However, these individuals' quotidian needs and issues are often neglected; they receive little aid for their day to day or long term functions. Two daily problems visually handicapped persons experience are trouble paying with bills - due to an inability to distinguish one amount from another - and difficulty eating at restaurants because they cannot locate their food or utensils (American Foundation of the Blind).

My application uses modern open source libraries including those utilizing artificial intelligence to distinguish bill amounts and locate objects. Object location is relayed to the user with a clockwise direction, ie. one o'clock, two o'clock, etc. Forums contributed to by the blind community mention how beneficial these clockwise directions are for finding objects. Various libraries and resources were instituted to help the visually impaired operate their everyday tasks with more ease.

The Bill Identification aspect utilizes artificial intelligence technology to recognize the text off of pictures of paper currency. Trained with font similar to the federal font used in bills, the accuracy of the bill identification improves with each use and correction of the algorithm.

The Location Detection feature integrates OpenCV, the open source computer vision library, into XCode, specifically the Features2D and Homography components. The algorithm detects key points in the object and locates those points given a different picture with the object in the background. As the computation involves points rather than areas in the picture, rotation and other transformations of the object do not deter the program. Objects that are irregularly shaped and unique to their design, such as wallets, are located by attaching a sticker of a registered and detectable logo. The code returns the location of the logo detected and computes the angle it is transformed from the meridian of the image using basic geometry. From the angle, the application calculates the clockwise orientation and notifies the user.

To communicate with the visually impaired, Apple's voice feedback toolkit Sirikit was utilized. The Siri function of iOS devices allows the blind to navigate through the application's functions using their voice alone. The command “open eyeHelper” launches the app; “open eyeHelper Bill Identification" commences the bill identification use, and “open eyeHelper Locator” beings the object identification process.

The application offers two functions, bill identification and object detection, to assist the blind community. With technology's increasing progression, the visually impaired should not be disregarded; technology should develop to calibrate to the needs of the disadvantaged. There are many improvements that could be made with my current model; the application could include an external camera, perhaps on glasses, that takes photos regularly to eradicate the need to aim and point the phone camera. Haptic instead of vocal feedback could be used to transmit messages to the visually impaired user to prevent the disruption of conversations or other vocal and auditory activities. Moreover, an Android application could be developed to further serve a larger market, eyeHelper serves as an efficient and cost-effective tool to aid the many visually handicapped individuals worldwide.