Tutorial for feature matching

The first step is to train your pattern set. The pattern object contains a category (in this case it will be "matching"), an image file, or url to image file, and a label. The image file can be provided passing the parameter "file", but if the image file is passed as a URL, use the parameter "remote_file_url". The label is the field used to store a unique ID. If the matching request returns values, the label is used to find more relevant content on your own database.

Request
Response

The second step is to force the update of indexes for matching. This step is optional and only used when you need to test the functionality. The indexes are updated with a delay of 60 to 120 seconds. This issue will be fixed in the next releases.

Request
Response

The third step is to test the feature matcher endpoint. In this case we will provide the image as a file. But you can pass the image via URL using the parameter "remote_file_url".

Request
Response