This GitHub repository provides an implementation of image label detection using Google Cloud Vision API, specifically for single object detection and multiple object detection. It also includes an example of training a custom AI model using Google Cloud AutoML Vision to improve object recognition accuracy.
Image label detection is a common task in computer vision applications, where the goal is to automatically identify and classify objects present in an image. This repository offers a comprehensive solution for image label detection using the powerful Cloud Vision API and demonstrates the utilization of Cloud AutoML Vision for training custom models.
The repository is organized into the following sections:
-
Single Object Detection: This section focuses on detecting a single object in an image. It provides a script that leverages the Cloud Vision API to process an input image and return the label or labels associated with the most prominent object detected. The script handles image pre-processing, API request/response handling, and label extraction.
-
Multiple Object Detection: This section extends the previous approach to handle scenarios where multiple objects may be present in an image. It utilizes the Cloud Vision API's capabilities to detect and label multiple objects simultaneously. The provided script takes an input image, processes it using the Cloud Vision API, and retrieves the labels associated with each detected object.
-
Cloud AutoML Vision: This section demonstrates the power of training custom AI models using Google Cloud AutoML Vision. It provides an example workflow for creating and training a custom model to improve object recognition accuracy. The repository includes a sample dataset, along with detailed instructions on training the model using the AutoML Vision service.
-
Easy Integration: The provided code and scripts can be easily integrated into existing projects or used as a starting point for new applications requiring image label detection.
-
Single and Multiple Object Detection: The repository includes separate implementations for single and multiple object detection, allowing users to choose the appropriate approach based on their specific needs.
-
Custom Model Training: The inclusion of the Cloud AutoML Vision section demonstrates how to train custom models using Google's AutoML service, providing users with the flexibility to create specialized models tailored to their specific domains or object categories.
-
Example Dataset: The repository includes a sample dataset that can be used to train a custom AI model using Cloud AutoML Vision. This dataset serves as a starting point for users who want to explore custom model training capabilities.
Contributions to this repository are welcome. If you find any bugs, have feature requests, or want to contribute improvements, please submit an issue or pull request following the guidelines outlined in the repository's CONTRIBUTING.md file.