@Ashok-Kumar-dharanikota can you tell me more details about the project.
seems very interesting but I really didn't understand how you are going to implement it

Hi @adithya-s-k ,

Thank you for responding to my issue. Let me take a simple scenario to explain this. for example, if we have had a code for a highly trained and highly accurate skin disease machine learning model. It would be great if we can make that model easily get integrated into mobile and web applications. That helps the individual or the person who's having a skin disease to check what type of skin disease he/she had and in which stage it is. We are bringing this high-level technology to people's pockets to help them utlise technology much better.

Here are a few examples to illustrate the benefits:

Real-time Object Detection: Converting a deep learning model for object detection into TensorFlow or PyTorch allows you to integrate it into a mobile app. Users can then perform real-time object detection on their devices, without relying on cloud services or internet connectivity.

Image Classification: By converting an image classification model to TensorFlow or PyTorch, you can create a mobile app that can classify images taken by the device's camera. The app can provide instant feedback to users, helping them identify objects, and landmarks, or even analyze medical images on their own devices.

Speech Recognition: Converting a speech recognition model to TensorFlow or PyTorch enables the deployment of voice-controlled applications on mobile devices. Users can interact with the app by speaking commands, which are then processed locally without the need for a network connection.

In summary, converting machine learning models into TensorFlow or PyTorch for edge devices brings efficiency, mobile compatibility, flexibility, offline functionality, and on-device inference. These benefits enable developers to create powerful, responsive, and user-friendly mobile applications that leverage the capabilities of machine learning while keeping data privacy intact.

really interesting

you can get started with the project
let me know how much time and as what rate you will make progress

you can start off in a fresh folder in the root directory of the repository

Hi @adithya-s-k

At the end, we will get three files .

• Android APK file
• Pyinb file
• tensorflow lite etc that can be used to integrate into mobile devices.

Do you want me to share all the files in a single folder. I will keep notifying you about the project. So you can decide the required files that can be added to the repo.

Hi @adithya-s-k ,

during my internship period at IIT Kharagpur, I worked on these ML/DL concepts. Initially, I created a Machine Learning model for Digit Classification and deployed it into the Application, and made On-Device offline Inference. I want to add this basic project to the repo first. Because it won't take much time and will give you a better idea about my Concept there are multiple videos on YT about this, but max trained with 10-20 epochs only I want to train with 50 to 100+ epochs so the model will give best accurate results. later I move to Skin, Heart, and Plant Disease classification and other Image, speech, and Object Classification concepts. Truly Appreciate your Feedback.

Thank you.

Hi @adithya-s-k ,

I did my work. I created a separate Folder for On-Device Machine Learning for Edge Devices. In this, I added three files to the new repo and separate readme.md file that introduces What is On-Device ML.

• App Source Code
• Tensorflow lite model
• Machine Learning file (.ipynb)

It would be great if you take a look at it the full folder.

https://github.com/Ashok-Kumar-dharanikota/World-of-AI

If you like my idea and Okay with the new project I would like to create a pull request. Please let me know your response.

Thank You.