February 2023 Summaries

The Clarifai Community has introduced new features and enhancements, including the Organizations feature, which allows users to consolidate multiple accounts for improved collaboration and centralized management. Additionally, several models and templates have been published, such as the "Recommended" Resnet template for optimal visual classification results, and moderation models for identifying inappropriate content and detecting cyberbullying in Japanese and Korean texts. The platform has also made improvements to the user interface and fixed numerous bugs related to app duplication, model loading, and API requests, among others. Users are encouraged to join the Clarifai Slack Group to engage with other users and access support from Clarifai engineers.

A detailed walkthrough describes how to perform image prediction using Clarifai's Python API, presented in a Jupyter notebook available on GitHub. The process involves installing necessary software and dependencies, including Clarifai's gRPC client and Protocol Buffers (protobuf) for communicating with Clarifai's servers. Users are guided through setting up authorization with Personal Access Tokens (PATs) for accessing Clarifai services. The tutorial explains how to initialize the Clarifai client, prepare image inputs as byte streams, and utilize Clarifai's General Model for visual classification. The model is capable of identifying various concepts in images and videos, and the results are visualized by plotting diagrams comparing identified concepts with their probabilities. The tutorial encourages users to modify the notebook to test their own images and classes, offering community support through a Slack channel for any questions.