How to use an AI Model and Video File of your own

For docker compose based deployment

You can bring your own model and run this sample application the same way as how we bring in the pallet defect detection model. You can also bring your own video file source. Please see below for details:

1. The pallet defect detection model is placed as below in the repository under models. You can also find the input video file source for inference under videos.

• resources/

  • models/

    • geti/

      • pallet_defect_detection/

        • deployment/

          • Detection/

            • model/

              • model.bin

              • model.xml

  • videos/

    • warehouse.avi

  Note
  You can organize the directory structure for models for different use cases.

2. The resources folder containing both the model and video file is volume mounted into EVAM in docker-compose.yml (present in the repository) file as follows:

   volumes:
   - ./resources/:/home/pipeline-server/resources/
   

3. Since this is a detection model, ensure to use gvadetect in the pipeline. For example: See the pallet_defect_detection pipeline in evam_config.json (present in the repository) where gvadetect is used.

4. The evam_config.json is volume mounted into EVAM in docker-compose.yml as follows:

   volumes:
   - ./configs/evam_config.json:/home/pipeline-server/config.json
   

5. Provide the model path and video file path in the REST/curl command for starting an inferencing workload. Example:

       curl http://<HOST_IP>:8080/pipelines/user_defined_pipelines/pallet_defect_detection -X POST -H 'Content-Type: application/json' -d '{
           "source": {
               "uri": "file:///home/pipeline-server/resources/videos/warehouse.avi",
               "type": "uri"
           },
           "destination": {
               "frame": {
                   "type": "webrtc",
                   "peer-id": "samplestream"
               }
           },
           "parameters": {
               "detection-properties": {
                   "model": "/home/pipeline-server/resources/models/geti/pallet_defect_detection/deployment/Detection/model/model.xml",
                   "device": "CPU"
               }
           }
       }'
   

For helm chart based deployment

You can bring your own model and run this sample application the same way as how we bring in the pallet defect detection model as follows:

1. The pallet defect detection model is placed as below in the repository under models. You can also find the input video file source for inference under videos.

• resources/

  • models/

    • geti/

      • pallet_defect_detection/

        • deployment/

          • Detection/

            • model/

              • model.bin

              • model.xml

  • videos/

    • warehouse.avi

  Note
  You can organize the directory structure for models for different use cases.

2. Build EVAM by including your new AI model (pallet defect detection model used here as an example) and video file. Example steps below:

   • Create the below Dockerfile at the root of the repository

     FROM intel/edge-video-analytics-microservice:2.4.0
     
     # Copy the application files
     COPY ./resources/models/geti /home/pipeline-server/resources/models/geti
     COPY ./resources/videos/warehouse.avi /home/pipeline-server/resources/videos/warehouse.avi
     
     # Define the command to run the application
     ENTRYPOINT ["./run.sh"]
     

   • Build a new EVAM image with the below command from the root of the repository

     docker build -t intel/edge-video-analytics-microservice:2.4.0 .
     

   • Please update imagePullPolicy as imagePullPolicy: IfNotPresent in values.yaml in order to use the above built image.

3. Since this is a detection model, ensure to use gvadetect in the pipeline. For example: See the pallet_defect_detection pipeline in evam_config.json (present in the repository) where gvadetect is used.

4. The evam_config.json is volume mounted into EVAM in provision-configmap.yaml as follows:

   apiVersion: v1
   kind: ConfigMap
   metadata:
     namespace: {{ .Values.namespace }}
     name: evam-config-input
   data:
     config.json: |-
   {{ .Files.Get "evam_config.json" | indent 4 }}
   

5. Provide the model path and video file path in the REST/curl command for starting an inferencing workload. Example:

       curl http://<HOST_IP>:30107/pipelines/user_defined_pipelines/pallet_defect_detection -X POST -H 'Content-Type: application/json' -d '{
           "source": {
               "uri": "file:///home/pipeline-server/resources/videos/warehouse.avi",
               "type": "uri"
           },
           "destination": {
               "frame": {
                   "type": "webrtc",
                   "peer-id": "samplestream"
               }
           },
           "parameters": {
               "detection-properties": {
                       "model": "/home/pipeline-server/resources/models/geti/pallet_defect_detection/deployment/Detection/model/model.xml",
                       "device": "CPU"
               }
           }
       }'