Visual attribution is the problem of identifying object features using its photo. This problem is very important as it is often used as part of a solution for a wide range of tasks. Among these tasks can be:
- Making a well-attributed catalog which will give users the ability to filter product specifying their characteristics.
- Making recommendations showing similar products.
- Making visual search allowing to find products by images.
In this showcase, we will show what is a visual attribution solving this problem for women dresses. At the end of the path, you could upload an image with a dress and get a prediction for such attributes as length and color.
To train the model you need data. You can use this data which holds on Google Cloud Storage and is publicly available.
AWS
- SageMaker
- S3 buckets to store data
- ECS to deploy backend service with UI