The SageMaker Image Classification algorithm can apply multiple labels to an image depending on what objects are identified. Objects are either identified, or not, there are no probability scores.
- AWS docs: https://docs.aws.amazon.com/sagemaker/latest/dg/image-classification.html
- AWS blogs: https://aws.amazon.com/blogs/machine-learning/classify-your-own-images-using-amazon-sagemaker/
|Data types and format||Image|
|Learning paradigm or domain||Image Processing, Supervised|
|Problem type||Image and multi-label classification|
|Use case examples||Label/tag an image based on the content of the image|
The recommended format for training data for the image classification algorithm is the Apache MxNet recordIO format, although it will also accept jpeg and png. Incremental training can be used to speed up training. This is where a previously trained model is used for the training data.
Model artifacts and inference
|Request format||application/x-image, jpeg, png|
Only GPU instances are recommended for the Image Classification algorithm. Multiple GPU instances and single instances with multiple GPUs can be used.
AWS Innovate | Intro to Deep Learning: Building an Image Classifier on Amazon SageMaker
This is a one hour and four minute video featuring the Image Classification algorithm, by Gabe Hollombe, from AWS.