EVS client ingests video at a pre-specified frameRate defined in the deployment manifest. Counting results from Rocket are serialized as a JSON object, and sent back to EVS client container as response of the corresponding POST request.
- On your on-prem edge, open two ssh sessions, one for monitoring evs client logs and the other for observing Rocket video ML container logs. To split screens, you can use an application like tmux.
Print Logs using:
sudo iotedge logs rocket -f
sudo iotedge logs evsclient -f
- Once deployed, EVS client will start ingesting video and posting frames to Rocket container for Background Detection and Line Occupancy Analysis.
Image Post Received. Count: 1752 Image Size: 521.8KB 0720*0400 Time: 09:08:38.644 PM
Image Post Received. Count: 1753 Image Size: 527.1KB 0720*0400 Time: 09:08:38.852 PM
Image Post Received. Count: 1754 Image Size: 523.5KB 0720*0400 Time: 09:08:39.046 PM
Image Post Received. Count: 1755 Image Size: 522.7KB 0720*0400 Time: 09:08:39.252 PM
Image Post Received. Count: 1756 Image Size: 522.2KB 0720*0400 Time: 09:08:39.445 PM
Image Post Received. Count: 1757 Image Size: 527.7KB 0720*0400 Time: 09:08:39.646 PM
Image Post Received. Count: 1758 Image Size: 524.2KB 0720*0400 Time: 09:08:39.816 PM
Image Post Received. Count: 1759 Image Size: 524.6KB 0720*0400 Time: 09:08:40.011 PM
Image Post Received. Count: 1760 Image Size: 523.5KB 0720*0400 Time: 09:08:40.213 PM
Image Post Received. Count: 1761 Image Size: 527.8KB 0720*0400 Time: 09:08:40.412 PM
Image Post Received. Count: 1762 Image Size: 525.1KB 0720*0400 Time: 09:08:40.617 PM
Image Post Received. Count: 1763 Image Size: 525.9KB 0720*0400
Post Received. Image Size: 539165 Time: 10/29/2021 09:08:41.638 PM bSize: 1
VAPCore.RunOneFrame
{"inferences":[{"other":{"count":2,"inferenceTime":26.2259},"subtype":"statistics","type":"other"},{"event":{"name":"Left-entrance","properties":{"accumulated":5,"count":0}},"subtype":"lineCounting","type":"event"},{"event":{"name":"Right-exit","properties":{"accumulated":0,"count":0}},"subtype":"lineCounting","type":"event"}]}
sFactor: 1 rFactor: 1 FrameID: 1767 ProcessTime: 26.23 Avg. Processing FPS: 39.48 Avg. E2E FPS: 3.20
BGS counts: 5 Cheap DNN counts: 0 Heavy DNN counts: 0 Http DNN counts: 0
- Object detection - As soon as an object is spotted, a frame buffer is encoded as a video clip and sent to the evsclient container on the network edge (e.g., AKS).
The example below shows an object is confirmed by the remote evs client after a heavy DNN operation.
You may examine the remote EVS Client's response by running
kubectl logs <YOUR_EVS_POD_NAME> evs -fon your network edge cluster.
** object detected by local AI module.
Image Post Received. Count: 1578 Image Size: 505.9KB 0720*0400 Time: 09:08:16.082 PM
Image Post Received. Count: 1579 Image Size: 528.5KB 0720*0400 Time: 09:08:16.170 PM
Image Post Received. Count: 1580 Image Size: 536.7KB 0720*0400 Time: 09:08:16.290 PM
Image Post Received. Count: 1581 Image Size: 545.0KB 0720*0400 Time: 09:08:16.368 PM
Image Post Received. Count: 1582 Image Size: 548.6KB 0720*0400 Time: 09:08:16.479 PM
Image Post Received. Count: 1583 Image Size: 547.4KB 0720*0400 Time: 09:08:16.575 PM
Image Post Received. Count: 1584 Image Size: 548.9KB 0720*0400 Time: 09:08:16.704 PM
Image Post Received. Count: 1585 Image Size: 554.6KB 0720*0400 Time: 09:08:16.815 PM
Image Post Received. Count: 1586 Image Size: 551.8KB 0720*0400 Time: 09:08:16.941 PM
Sending encoded frame buffer (88.1KB) to a remote EVS client.
Image Post Received. Count: 1587 Image Size: 551.9KB 0720*0400 Time: 09:08:17.049 PM
Image Post Received. Count: 1588 Image Size: 538.8KB 0720*0400 Time: 09:08:17.155 PM
Image Post Received. Count: 1589 Image Size: 538.2KB 0720*0400 Time: 09:08:17.246 PM
Image Post Received. Count: 1590 Image Size: 539.4KB 0720*0400 Time: 09:08:17.336 PM
Image Post Received. Count: 1591 Image Size: 544.1KB 0720*0400 Time: 09:08:17.433 PM
Image Post Received. Count: 1592 Image Size: 543.3KB 0720*0400 Time: 09:08:17.528 PM
Image Post Received. Count: 1593 Image Size: 543.9KB 0720*0400 Time: 09:08:17.617 PM
Image Post Received. Count: 1594 Image Size: 544.4KB 0720*0400 Time: 09:08:17.712 PM
Image Post Received. Count: 1595 Image Size: 549.7KB 0720*0400 Time: 09:08:17.800 PM
Image Post Received. Count: 1596 Image Size: 550.5KB 0720*0400 Time: 09:08:17.892 PM
Image Post Received. Count: 1597 Image Size: 550.9KB 0720*0400 Time: 09:08:17.983 PM
Image Post Received. Count: 1598 Image Size: 553.1KB 0720*0400 Time: 09:08:18.075 PM
Result confirmed by a remote EVS client.
Response: {"inferences":[{"other":{"count":2,"frameid":-1,"inferenceTime":-1},"subtype":"statistics","type":"other"},{"event":{"name":"Left-entrance","properties":{"accumulated":5,"count":1}},"subtype":"lineCounting","type":"event"},{"event":{"name":"Right-exit","properties":{"accumulated":0,"count":0}},"subtype":"lineCounting","type":"event"}]}
You can view the output images inside /evsrocket/output_all on the on-prem edge device. Results from different modules are sent to different directories (e.g., output_bgsline for background subtraction-based detector) whereas output_all has images from all modules.
❗Note: /evsrocket/rocket/output_all will list all the frames captured by rocket after background subtraction.
❗Note: /evsrocket/evsclient will list all the buffers which was forwarded by local evs client to remove evs client for DNN Operations
In the mounted file share that we specified while setting AKS, you can see the output images processed by Network edge after a heavy DNN operation. Review the images in the evs/output folder/file share in your storage account.
Reference Output:
