add and activate point robinson & sunset bay hydrophones (#136)

* add and activate sunset-bay hydrophone

* add and activate point robinson hydrophone

InferenceSystem/README.md

@@ -11,16 +11,16 @@ Note: We use Python 3, specifically tested with Python 3.7.4
 # How to run the InferenceSystem locally
 ## Create a virtual environment
 
-1. In your working directory, run `python -m venv inference-venv`. This creates a directory `inference-venv` with relevant files/scripts. 
-2. On Mac, activate this environment with `source inference-venv/bin/activate` and when you're done, `deactivate`
+1. In your working directory, run `pip install virtualenv && virtualenv inference-venv`. This creates a directory `inference-venv` with relevant files/scripts. 
+2. On Mac or Linux, activate this environment with `source inference-venv/bin/activate` and when you're done, `deactivate`
 
     On Windows, activate with `.\inference-venv\Scripts\activate.bat` and `.\inference-venv\Scripts\deactivate.bat` when done
 3. In an active environment, cd to `/InferenceSystem` and run `python -m pip install --upgrade pip && pip install -r requirements.txt` 
 
 ## Model download
 
 1.  Download the current production model from [this link.](https://trainedproductionmodels.blob.core.windows.net/dnnmodel/11-15-20.FastAI.R1-12.zip)
-2.  Unzip *.zip and extract to `InferenceSystem/model`.
+2.  Unzip *.zip and extract to `InferenceSystem/model` using `unzip 11-15-20.FastAI.R1-12.zip`
 3.  Check the contents of `InferenceSystem/model`.
 There should be 1 file
     * model.pkl
@@ -149,8 +149,21 @@ This can be completed in two ways.
         ```
         AZURE_COSMOSDB_PRIMARY_KEY=<key>
         AZURE_STORAGE_CONNECTION_STRING=<string>
+        INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING=<string>
         ```
 
+## Adding a new hydrophone
+
+1. Create a new config file under the [config](config) folder
+
+2. Update the last line of the [Dockerfile](Dockerfile) to point to the new config file 
+
+3. Create a new deployment YAML under the [deploy](deploy) folder
+
+4. Update [src/LiveInferenceOrchestrator.py](src/LiveInferenceOrchestrator.py) and [src/globals.py](src/globals.py) to add variables for the new hydrophone location 
+
+5. Follow all other steps below until you update the kubernetes cluster with the new namespace
+
 ## Building the docker container for production
 
 From the `InferenceSystem` directory, run the following command.
@@ -163,7 +176,8 @@ docker build . -t live-inference-system -f ./Dockerfile
 
 Note: the config used in the Dockerfile is a Production config.
 
-TODO: fix. For now, you will have to manually create 3 different docker containers for the 3 hydrophone locations. Each time you will need to edit the Dockerfile and replace the config for each hydrophone location (OrcasoundLab, BushPoint, PortTownsend).
+TODO: fix. For now, you will have to manually create 5 different docker containers for the 5 hydrophone locations. Each time you will need to edit the Dockerfile and replace the config for each hydrophone location (OrcasoundLab, BushPoint, PortTownsend, Sunset Bay and Point Robinson).
+
 
 ## Running the docker container
 
@@ -201,23 +215,21 @@ This step pushes your local container to the Azure Container Registry (ACR).  If
 documentation is adapted from 
 [this tutorial](https://docs.microsoft.com/en-us/azure/container-instances/container-instances-tutorial-prepare-acr).
 
-Login to the shared azure directory from the Azure CLI.
+1. Login to the shared azure directory from the Azure CLI.
 
 ```
-az login --tenant ai4orcasoutlook.onmicrosoft.com
+az login --tenant adminorcasound.onmicrosoft.com
 ```
 
-We will be using the orcaconservancycr ACR in the LiveSRKWNotificationSystem Resource Group.
-
-Log in to the container registry.
+2. We will be using the orcaconservancycr ACR in the LiveSRKWNotificationSystem Resource Group. Log in to the container registry.
 
 ```
 az acr login --name orcaconservancycr
 ```
 
 You should receive something similar to `Login succeeded`.
 
-Tag your docker container with the version number. We use the following versioning scheme.
+3. Tag your docker container with the version number. We use the following versioning scheme.
 
 ```
 docker tag live-inference-system orcaconservancycr.azurecr.io/live-inference-system:<date-of-deployment>.<model-type>.<Rounds-trained-on>.<hydrophone-location>.v<Major>
@@ -230,7 +242,8 @@ docker tag live-inference-system orcaconservancycr.azurecr.io/live-inference-sys
 ```
 
 Look at [deploy-aci.yaml](deploy-aci.yaml) for examples of how previous models were tagged.
-Lastly, push your image to Azure Container Registry for each Orcasound Hydrophone Location.
+
+4. Lastly, push your image to Azure Container Registry for each Orcasound Hydrophone Location.
 
 ```
 docker push orcaconservancycr.azurecr.io/live-inference-system:<date-of-deployment>.<model-type>.<Rounds-trained-on>.<hydrophone-location>.v<Major>
@@ -284,7 +297,7 @@ kubectl create secret generic inference-system -n bush-point \
 4. Create or update deployment. Use file for hydrophone under [deploy](./deploy/) folder, or create and commit a new one.
 
 ```bash
-kubectl apply -f bush-point.yaml
+kubectl apply -f deploy/bush-point.yaml
 ```
 
 5. To verify that the container is running, check logs:
@@ -300,6 +313,7 @@ kubectl logs -n bush-point inference-system-6d4845c5bc-tfsbw
 <details>
   <summary>Deployment to Azure Container Instances (deprecated)</summary>
 # Deploying an updated docker build to Azure Container Instances
+# This method has been deprecated
 
 ## Prerequisites
 

InferenceSystem/config/Production/FastAI_LiveHLS_PointRobinson.yml

@@ -0,0 +1,10 @@
+model_type: "FastAI"
+model_local_threshold: 0.5
+model_global_threshold: 3
+model_path: "./model"
+model_name: "model.pkl"
+hls_stream_type: "LiveHLS"
+hls_polling_interval: 60
+hls_hydrophone_id: "rpi_point_robinson"
+upload_to_azure: True
+delete_local_wavs: True

InferenceSystem/config/Production/FastAI_LiveHLS_SunsetBay.yml

@@ -0,0 +1,10 @@
+model_type: "FastAI"
+model_local_threshold: 0.5
+model_global_threshold: 3
+model_path: "./model"
+model_name: "model.pkl"
+hls_stream_type: "LiveHLS"
+hls_polling_interval: 60
+hls_hydrophone_id: "rpi_sunset_bay"
+upload_to_azure: True
+delete_local_wavs: True

InferenceSystem/demos/hls_reader.py

@@ -21,6 +21,7 @@
     # 'OrcasoundLab': 'https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_orcasound_lab'
     'BushPoint': 'https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_bush_point'
     # 'PortTownsend': 'https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_port_townsend'
+    'SunsetBay': 'https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_sunset_bay'
 }
 
 dirname = os.getcwd()

InferenceSystem/deploy-aci.yaml

@@ -50,6 +50,36 @@ properties:
         requests:
           cpu: 1.0
           memoryInGB: 3
+  - name: sunset-bay-live
+    properties:
+      environmentVariables:
+      - name: AZURE_COSMOSDB_PRIMARY_KEY
+        secureValue: '<cosmos_primary_key>'
+      - name: AZURE_STORAGE_CONNECTION_STRING
+        secureValue: '<storage_connection_string>'
+      - name: INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING
+        secureValue: '<appinsights_connection_string>'
+      image: orcaconservancycr.azurecr.io/live-inference-system:11-15-20.FastAI.R1-12.SunsetBay.v0
+      ports: []
+      resources:
+        requests:
+          cpu: 1.0
+          memoryInGB: 3
+  - name: point-robinson-live
+    properties:
+      environmentVariables:
+      - name: AZURE_COSMOSDB_PRIMARY_KEY
+        secureValue: '<cosmos_primary_key>'
+      - name: AZURE_STORAGE_CONNECTION_STRING
+        secureValue: '<storage_connection_string>'
+      - name: INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING
+        secureValue: '<appinsights_connection_string>'
+      image: orcaconservancycr.azurecr.io/live-inference-system:11-15-20.FastAI.R1-12.PointRobinson.v0
+      ports: []
+      resources:
+        requests:
+          cpu: 1.0
+          memoryInGB: 3
   imageRegistryCredentials:
   - server: orcaconservancycr.azurecr.io
     username: orcaconservancycr

InferenceSystem/deploy/point-robinson.yaml

@@ -0,0 +1,38 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: inference-system
+  namespace: point-robinson
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: inference-system
+  template:
+    metadata:
+      labels:
+        app: inference-system
+    spec:
+      containers:
+      - name: inference-system
+        image: orcaconservancycr.azurecr.io/live-inference-system:11-15-20.FastAI.R1-12.PointRobinson.v0
+        resources:
+          limits:
+            cpu: 1
+            memory: 3G
+        env:
+          - name: AZURE_COSMOSDB_PRIMARY_KEY
+            valueFrom:
+              secretKeyRef:
+                name: inference-system
+                key: AZURE_COSMOSDB_PRIMARY_KEY
+          - name: AZURE_STORAGE_CONNECTION_STRING
+            valueFrom:
+              secretKeyRef:
+                name: inference-system
+                key: AZURE_STORAGE_CONNECTION_STRING
+          - name: INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING
+            valueFrom:
+              secretKeyRef:
+                name: inference-system
+                key: INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING

InferenceSystem/deploy/sunset-bay.yaml

@@ -0,0 +1,38 @@
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: inference-system
+  namespace: sunset-bay
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: inference-system
+  template:
+    metadata:
+      labels:
+        app: inference-system
+    spec:
+      containers:
+      - name: inference-system
+        image: orcaconservancycr.azurecr.io/live-inference-system:11-15-20.FastAI.R1-12.SunsetBay.v0
+        resources:
+          limits:
+            cpu: 1
+            memory: 3G
+        env:
+          - name: AZURE_COSMOSDB_PRIMARY_KEY
+            valueFrom:
+              secretKeyRef:
+                name: inference-system
+                key: AZURE_COSMOSDB_PRIMARY_KEY
+          - name: AZURE_STORAGE_CONNECTION_STRING
+            valueFrom:
+              secretKeyRef:
+                name: inference-system
+                key: AZURE_STORAGE_CONNECTION_STRING
+          - name: INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING
+            valueFrom:
+              secretKeyRef:
+                name: inference-system
+                key: INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING

InferenceSystem/requirements.txt

@@ -24,6 +24,5 @@ numba==0.48
 opencv-python
 boto3
 pytz
-json
 opencensus-ext-azure
 orca-hls-utils

InferenceSystem/src/LiveInferenceOrchestrator.py

@@ -21,7 +21,7 @@
 from azure.cosmos import exceptions, CosmosClient, PartitionKey
 
 import sys
-
+import logging
 from opencensus.ext.azure.log_exporter import AzureLogHandler
 from opencensus.ext.azure.log_exporter import AzureEventHandler
 
@@ -36,8 +36,10 @@
 ORCASOUND_LAB_LOCATION = {"id": "rpi_orcasound_lab", "name": "Haro Strait", "longitude":  -123.17357, "latitude": 48.55833}
 PORT_TOWNSEND_LOCATION = {"id": "rpi_port_townsend", "name": "Port Townsend", "longitude":  -122.76045, "latitude": 48.13569}
 BUSH_POINT_LOCATION = {"id": "rpi_bush_point", "name": "Bush Point", "longitude":  -122.6039, "latitude": 48.03371}
+SUNSET_BAY_LOCATION = {"id": "rpi_sunset_bay", "name": "Sunset Bay", "longitude":  -122.3339, "latitude": 47.86497}
+POINT_ROBINSON_LOCATION = {"id": "rpi_point_robinson", "name": "Point Robinson", "longitude":  -122.37267, "latitude": 47.38838}
 
-source_guid_to_location = {"rpi_orcasound_lab" : ORCASOUND_LAB_LOCATION, "rpi_port_townsend" : PORT_TOWNSEND_LOCATION, "rpi_bush_point": BUSH_POINT_LOCATION}
+source_guid_to_location = {"rpi_orcasound_lab" : ORCASOUND_LAB_LOCATION, "rpi_port_townsend" : PORT_TOWNSEND_LOCATION, "rpi_bush_point": BUSH_POINT_LOCATION, "rpi_sunset_bay": SUNSET_BAY_LOCATION, "rpi_point_robinson": POINT_ROBINSON_LOCATION }
 
 def assemble_blob_uri(container_name, item_name):
 
@@ -99,11 +101,11 @@ def populate_metadata_json(
 		config_params = yaml.load(f, Loader=yaml.FullLoader)
 
 	# logger to app insights
-	con_string = os.getenv('INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING')
+	app_insights_connection_string = os.getenv('INFERENCESYSTEM_APPINSIGHTS_CONNECTION_STRING')
 	logger = logging.getLogger(__name__)
-	if appInsightsKey is not None:
-		logger.addHandler(AzureLogHandler(connection_string=con_string))
-		logger.addHandler(AzureEventHandler(connection_string=con_string))
+	if app_insights_connection_string is not None:
+		logger.addHandler(AzureLogHandler(connection_string=app_insights_connection_string))
+		logger.addHandler(AzureEventHandler(connection_string=app_insights_connection_string))
 		logger.setLevel(logging.INFO)
 
 	## Model Details

InferenceSystem/src/globals.py

@@ -4,6 +4,8 @@
     "orcasound_lab": "https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_orcasound_lab",
     "port_townsend": "https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_port_townsend",
     "bush_point": "https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_bush_point",
+    "sunset_bay": "https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_sunset_bay",
+    "point_robinson": "https://s3-us-west-2.amazonaws.com/streaming-orcasound-net/rpi_point_robinson",
 }
 
 # Limits time window (end - start) of negative samples to be downloaded for retraining