Auto Tagging
Image Auto Tagging technology enables automatic assignment of relevant tags or keywords to vast collections of images and videos.
![Auto image tagging visual explanation](/sites/default/files/styles/16_9_100/public/2024-07/auto-tagging-explanation.jpg?itok=53SNqASE)
For other use, please contact info@imagga.com
Image Auto tagging is a core capability of Imagga’s computer vision offering. The image tagging deep learning model analyses the pixel content of visuals, extracts their features and detects objects of interest. The model is trained with more than 3000 objects from daily life. It is a cost-effective and time-saving solution when operating with massive amounts of image content often coming from different sources and with incoherent meta-data or even worse – lack of any metadata. Image tagging enables classification of visual data through the use of tags and labels. This allows for the quick searching and identifying of images, as well as the adequate categorisation of visuals in the databases.
Hardware architecture:
X64
Install & Run:
The docker image contains a runnable server allowing Imagga Image Auto Tagging to be used through gRPC.
We have also provided two files in a zip archive:
- model.proto - a Protocol Buffers (protobuf) file that defines the data structures (messages) and services (RPC methods) used for communication between clients and servers for the Image auto tagging service
- test_client.py - a python sample script calling the tagging service
1. First you run the docker container Imagga AI4Media
Example pull and run:
$ docker pull alejandro131/imagga-ai4media
$ docker run -p 8061:8061 --name ai4media_imagga_tagging <docker_image_id>
Note: Take into account that the server is running on port 8061
2. Then you need to generate a few additional files for gRPC based on model.proto file with gRPC tools:
- model_pb2.py & model_pb2.pyi - it contains the generated request and response classes
- Model_pb2_grpc.py - it contains the generated client and server classes
In order to generate them, you need gRPC and gRPC tools:
$ python -m pip install grpcio
$ python -m pip install grpcio-tools
Then you can call this command:
$ python -m grpc_tools.protoc -I./ --python_out=. --pyi_out=. --grpc_python_out=. /model.proto
Here a few links with a detailed explanation of the gRPC and a sample with the python language: gRPC docs and gRPC for Python.
3. Finally you can test by running the python script test_client.py. It shows how to use the Imagga Tagging with a gRPC call.
Note that you have to use IMAGGA KEY and SECRET to be able to perform the tagging inference.
To acquire the api credentials you should sign up to the Imagga site Imagga Sign Up.