The image classification model generates images descriptors to classify the subject of an image.
The model deploys the ResNet-50 v1.5 model from Hugging Face. It requires a JPEG, JPG, or PNG image file that is a maximum of 5 MB as input. The model returns five classifications with a confidence score for each classification.
The results from the image classification model might differ from the results you see on Hugging Face for the ResNet-50 v1.5 model. Hugging Face sometimes changes the model on their website, while the image classification model is static.
Note: This model is sourced from a third party. Informatica does not assume ownership of its behavior, functionality, or any associated liabilities. Any issues, updates, or concerns related to this artifact should be directed to the original provider or creator. We provide this artifact as-is and without warranties.
Example
You run an online store that receives customer photos in reviews. Before the photos are posted to your website, you need to check that they're relevant to the product being reviewed. As an initial check, you use the image classification model to classify the photos. If a photo returns the right product type with a confidence score over 0.5, you can publish the review. Otherwise, you flag the photo to be reviewed by a content moderator.
The following image shows an example test of the image classification model:
