ML-Kit is a free-to-use machine learning API for Android and
iOS apps, focusing
on on-device AI solutions.
In this repo, I have made a demo of all the AI techniques I have performed with ML-Kit on Android.
The ML techniques used in this demo app include but are not limited to:
- This algorithm detects face mesh info on close-range images.
- This algorithm can recognize and extract text from images.
- Detecting the position of human body in any given picture/video in real-time.
- Separates the background of a picture/video from users within it. Helps to focus on more important objects in the picture/video.
- Localize and tag in realtime one or more objects in the live camera feed.
- Scanning and processing most kinds of barcodes. Supports various standard 1D and 2D (a.k.a. QR) barcode formats.
- This algorithm identifies objects, locations, activities, animal species, products and more in a given picture.
- Detects faces and facial landmarks in a given image/video.
- This amazing demo camera app firstly determines the language of a string of text with just a few words. And then, translates that text between 58 languages; completely on device.
- This part of the app recognizes handwritten text and hand-drawn shapes (such as emojis) on a digital surface, such as a touch screen.
Example results:

Some examples:


Example result:



Example results of object detection:





Example result of barcode scanner:

A video demo of barcode scanner:
Realtime barcode scanner video demo
barcode_demo.mp4
For example, in the picture below, it has managed to label the road, Jeans, Jacket, and Buildings in the picture correctly.

Example results:

A video demo of realtime camera translator:
Live camera translator video demo
live.camera.translator.mp4
Some examples:


A video demo of digital ink recognition:
Digital ink recognition video demo
2023-04-11.01.05.42.mp4
Please star and fork this repo, I will be maintaining it over time and will try to add more ML
related library demos to
it.
Feel free to open issues and point out the bugs and short-comings.
new architecture of the app is going to be like this: we will have these 5 main modules in our codebase:
- feature_ink_recognition
- feature_live_translator
- feature_still_image
- feature_live_preview
- shared