Amazon Rekognition Custom Labels

Marketing agencies need to accurately report on brand coverage of their clients in various media. Typically they manually track appearances of their clients’ logos and products in social media images, broadcast, and sports videos. With Amazon Rekognition Custom Labels, agencies can create a custom model specifically trained to detect their client logos and products. Instead of painstakingly trying to follow traditional and social media manually, they can process images and video frames through the custom model to find the number of impressions.

Content producers typically have to search through thousands of images and videos to find the relevant content they want to use for producing shows. For example, a sports broadcaster often needs to assemble highlight films about games, teams, and players for affiliates, which can take hours to manually assemble from archives. By training custom models to identify teams and players by jersey and number, and to identify common game events like goals scored, penalties, and injuries, they can quickly develop a relevant list of images and clips that match the subject of the film.

Agriculture companies need to rate the quality of their produce before packing them. For example, a tomato producer may manually classify tomatoes into 6 ripeness groups from mature green to red, and packs them accordingly to ensure maximum shelf life. Instead of manually examining each tomato, they can train a custom model to classify tomatoes based on their ripeness criteria. By integrating the model with their manufacturing systems, they can automatically sort the tomatoes, and pack them accordingly.

The Rekognition Custom Labels console provides a visual interface to make labeling your images fast and simple. The interface allows you to apply a label to the entire image or to identify and label specific objects in images using bounding boxes with a simple click-and-drag interface.

Alternately, if you have a large data set, you can use Amazon SageMaker Ground Truth to efficiently label your images at scale.

No machine learning expertise is required to build your custom model. Rekognition Custom Labels includes AutoML capabilities that take care of the machine learning for you. Once the training images are provided, Rekognition Custom Labels can automatically load and inspect the data, select the right machine learning algorithms, train a model, and provide model performance metrics.

Evaluate your custom model’s performance on your test set. For every image in the test set, you can see the side by side comparison of the model’s prediction vs. the label assigned. You can also review detailed performance metrics such as precision/recall metrics, f-score, and confidence scores. You can start using your model immediately for image analysis, or iterate and re-train new versions with more images to improve performance. After you start using your model, you track your predictions, correct any mistakes and use the feedback data to retrain new model versions and improve performance.