Objects classification service which provides simple REST API for images upload and search images by a class. Free tensorflow model published under the Apache license is used for the classification. H2 database is used as a data storagre.

1. Build JAR using mvn package command.
2. Create models directory and place tensorflow_inception_graph.* there.
3. Create public directory as a storage for the uploaded images.
4. Create application.properties and set needed parameters for the application.
5. Run the application using java -jar objclassifier-0.0.1-SNAPSHOT.jar command.

The project uses Spring boot, Spring Web and Spring data JPA, so you shoud use standard properties if you need to customize database or service.

You should set processing properties:

• processing.path

  Path to processed images storage. By default it is 'public'.

• processing.model

  Path to the tensorflow classification model.

• processing.groups

  Path to the groups list file.

Return a list of images groups.

Response example:

[
  "zebra",
  "African elephant",
  "tabby"
]

Return a list of images in a group {name}.

Response example:

[
  "1661588051129.jpg"
  "1661588055135.jpg"
]

Return a list of groups for specified image {name}

Response example:

[
  "zebra"
]

Process and classify image from file field of the request. File will be saved to the processing.path folder with name generated from unix epoch time.

Get a previously uploaded image with the name {image}.

You can do whatever you need with the code. Data and software should be totally free!