Atheia: Employing Redis in Developing a Visual Aid for Real-time Object Search

Visual aid technologies are meant for users to be more independent, however, the high price tag, reliability, and in-accessibility to active users prevented any inclination for users to obtain such products; devices such as Orcam were too expensive ($5900) containing limited functionality and long latency times. To address these issues, we created Atheia: a compact and affordable visual aid that allows for a user to fully and safely understand their environment with low latency.

Atheia is a glove that serves as a visual aid for the blind and works concurrently with a mobile app. The glove has a camera and a microcontroller embedded in the back of the hand. To utilize the device, the user has the back of their hand towards their surroundings and activates the button on the side of the glove to observe their surroundings. Using Redis Bit Fields and Strings, the image and appropriate user data is sent to an VPC instance. There we use several AWS models and services, and store a list of the objects surrounding the user, any text, and the location of each object in the environment. This data is also stored in a log so that the user can search for it in a respective feature. Using this data, the user can choose one of our five main features: Scan (describing the objects surrounding the user), Text (reading out any text found in the image), Observations (takes questions that look for details in an image), Search (guides users towards the desired objects which they choose through a system of vibrations), and Last Seen (returns the time, place, and location an object was last seen). To activate a feature, the user just has to give an audio command.

Because we run multiple models in parallel, we use Redis to communicate between the scripts since they run independently of each other. We used the Redis database to store our variables across our 3 interfaces, the microcontroller, the app, and the server. The entire process begins when the user clicks a button on the side of the glove. As soon as that occurs, an image is captured and uploaded to the Redis database through a byte field. When the image finishes uploading, it changes a Redis variable, called “read”, which triggers all the models to process the images. To manage running several computer vision models simultaneously and efficiently, we utilized Redis’s unique "enque" service and parallel processing. The queue contained jobs that included the AI services and models (Yolo5x, Detectron, EasyOCR, Pythia), and were only processed when a state change in the “read” variable occurred. Thanks to Redis’s queue service, we were able to simplify the process needed to manage all our machine learning services. With these technologies, we were able to create an affordable device and assist one of our most resilient communities: the visually impaired.