A collection of AI projects for mobile devices (Awesome Tensorflow Lite)

Tensorflow Lite is a software package that contains tools that allow AI models trained with Tensorflow to run on mobile devices. It is said to be running on over 4 billion devices now. Basically, since the trained model is converted to Tensorflow Lite format and used on a mobile device, it has the advantage of using the existing models as it is.

The results of many AI projects require high-performance GPUs and are made under the assumption that they operate in a PC environment, so there are many cases where a separate inference server is set up and accessed through API for service. However, in this case, the infrastructure cost increases proportionally as the number of users increases, and there is a problem that fundamental network latency is required. There is also an issue of utilization of personal information. This problem can be avoided by making inferences about the AI model directly on the user's device, and the utilization of on-device AI is expected to continue to increase, with a separate AI inference accelerator installed on smartphones recently.

A site called Awesome Tensorflow Lite is a collection of various mobile app sample projects using Tensorflow Lite. The following is a collection of projects organized in the awesome tensorflow lite github repository. There are projects in various fields such as compute vision, text, speech, and recommendation.

Computer vision

Text

Speech

Recommendation

There are quite a variety of examples, so it will be useful when developing a smartphone app that is directly utilized or applied with an AI model that has been directly researched and developed. Here is a link to the awesome tensorflow lite site repository:

margaretmz/awesome-tensorflow-lite

TensorFlow Lite models, samples, tutorials, tools and learning resources. – Margaretmz/awesome-tensorflow-lite