This is a re-upload as I consolidate all my AI-related videos into one channel. This is the video introduction I made as part of my submission requirements for a google tensorflow related hackathon at this link
https://devpost.com/software/radlense2


This video features RadLens2 which is the older version of

RadLens-AI.


RadLens-AI is featured in this you tube video
https://radhorizon.com/RadLens%20AI%20Foot%20Ossicles/





FROM the original RadLens2 video description:

RadLens2 has just been updated and migrated to the latest Tensorflow 2.0.

Check it out now at ....
https://radhorizon.com/radlens2/

Web app changes:
--now uses the native tf.keras api.
--smaller AI/ML model foot print after conversion from native tf python to tf js
--results in faster initial load times (10-30 seconds on some moderate internet connection speeds versus the older web app which used to take 1-2 mins for the initial load time).

RadLens2 is a web app powered by Artificial Intelligence/Machine Learning. This is an on going project I'm working on that can hopefully spread the word on how AI/ML is now becoming more and more accessible given powerful machine learning libraries like tensorflow which has just become even more straight forward to use given its adoption of Keras as a higher level API.

I'm a doctor/radiologist by profession. I hope I can convince more of my fellow rads to actively participate in the research, development and utilization of AI/ML in medicine.

The app tries to classify two fracture types: Galeazzi and Monteggia fractures of the forearm.

No images are uploaded to any server running in the background (keeps private date on the local machine). The AI/ML model/algorithm runs in real time on the local machine (phone or laptop) right in the browser (no need for special hardware). The app uses the camera to scan images and gives a prediction that then leads to a link for a google image search.

In the video, the app runs on an android phone (version 8.0/Oreo/Asus zenphone MaxPro M1/latest chrome).

I no longer used Keras as a wrapper since tensorflow 2.0 now natively supports the Keras api as a higher level api under tf.keras.
More power to the awesome devs behind these huge bodies of underlying code. Let's all #democratizeAI