Web application built with TypeScript, React, and the TensorFlow.js MobileNet API.
Taking into consideration that the TensorFlow model has only been trained with however many images, the name of this app is surprisingly accurate to what it's capable of! Like a cat's vision, this app will help with identifying a variety of images; however, it will not always be 100% accurate!
As a result, the aim of this web app was to provide a cozy environment for user's to upload images either through their computers or by pasting an image url in order to receive assistance with classifying or identifying their desired image! Often times than not, we have images of perhaps animals that we simply cannot identify. And so, MeowVision offers itself as a solution to assisting those in need of identifying objects in particular images!
- Uploading and pasting image urls for image classification
- "Drawer" for storage of most recently classified images for ease of retrieval
- Integrated the use of the TensorFlow.js MobileNet API for image classification of up to 3 AI guesses
- Responsive and lively user interface design
- Redesign the "drawer" feature to store images in a more aspect ratio friendly manner
- Incorporate a text-to-speech voice API to increase accessibility of the image recognition web app