Skip to content


This is a simple demo of DJL with AWS KVS. The demo runs object detection on each frame in the video.


To run the App demo, first create a KVS stream on an AWS account. Edit and set the REGION and STREAM_NAME to your created stream.

The App demo only retrieves from the stream, so you will also need to separately put media to the stream as well. You can use the PutMediaDemo from the Kinesis Java SDK. Edit the Demo to use the correct region and stream as well. You may also want to edit the video as well, because the default video does not detect any objects. You can find some additional videos here. Then, follow the instructions in the README to setup credentials and put to the stream.

Once data is in KVS, you can then run the main function in App. In order for App to work, you need to follow the same credential setup you used for the PutMediaDemo (see here).

The demo will print out what objects are detected through standard out. It will also save annotated frames in the out directory with boxes drawn over all detected objects.


You can also look at this demo which runs object detection for each frame in a file. Before running the demo, edit the FILENAME to the path of the video you want to run on. Then, it will output the detection on each frame and output to the out directory like the main App.