Globally, over 35 Google developer communities have hosted ML campaigns distributed by the ML developer programs team during the first half of 2023 showcasing their work and interests. Out of the many, we’ve picked out our favourite contributions by the Indian community.
Here are the top 8 highlights from Indian Machine Learning Google Developer Experts.
Image Segmentation Using Composable Fully-Convolutional Networks
In a demo, Suvaditya Mukherjee provided a detailed account of implementing a fully-convolutional network using a VGG-16 backend and its application in image segmentation.
Furthermore, in the KerasCV for the Young and Restless session he delved into the fundamental computer vision components. The presentation showed the significance of this tool and highlighted its integration with TFX and Keras ecosystem.
[ML Story] My Keras Chronicles
Aritra Roy Gosthipaty shared his journey into the world of deep learning, specifically with Keras. Within his narrative, he offered insights on how aspiring individuals could become a part of the open source community. Alongside him, Subvaditya presented their Keras implementation of Temporal Latent Bottleneck Networks, as originally proposed in the paper.
TensorFlow and Keras Implementation of the CVPR 2023 Paper
Usha Rengaraju showcased a research paper’s practical application, bringing to light the implementation of BiFormer – a novel Vision Transformer enriched with Bi-Level Routing Attention.
Skeleton Based Action Recognition: A Failed Attempt
Ayush Thakur provided insights into his experience of participating in the Kaggle competition, “Google – Isolated Sign Language Recognition“. He shared his repository, training logs, and the various approaches he employed during the competition. Furthermore, Ayush’s enlightening article, “Keras Dense Layer: How to Use It Correctly,” delved into the dense layer in Keras and its practical applications.
Add Machine Learning to Your Android App
At Tech Talks for Educators, Pankaj Rai conducted a session attended by 700+ individuals about how to bring ML capabilities into Android applications, such as object detection and gesture recognition. During the session, he explained ML Kit, MediaPipe, and TF Lite tools, demonstrating their capabilities and instructing attendees on how to effectively use these resources.
Google’s Bard Can Write Code
Bhavesh Bhatt showed the coding capabilities of Bard, how to create a 2048 game with it, and how to add some basic features to the game. He also uploaded videos about LangChain in a playlist and introduced Google Cloud’s new course on Generative AI.
In a demonstration, Bhatt showcased the coding capabilities of Bard like how to create a 2048 game using it. Furthermore, he illustrated the process of adding features to enhance the game’s functionality. Alongside this, he shared videos on LangChain and Google Cloud’s new Generative AI course.
Open and Collaborative MLOps
In favour of the open-source community, Sayak Paul spoke about why open and collaboration are two important aspects of MLOps. He gave an overview of Hugging Face Hub and how well it integrates with TFX in MLOps workflows.
During a discussion, Paul spoke about the significance of openness and collaboration in the context of MLOps. He further outlined the merits of Hugging Face Hub, showcasing its integration with TFX.
Learning JAX in 2023: Part 3 — A Step-by-Step Guide to Training Your First Machine Learning Model with JAX
Aritra Roy Gosthipaty and Ritwik Raha together showed how JAX can train linear and nonlinear regression models and the usage of PyTrees library to train a multilayer perceptron model.
