Merge pull request #73 from FR-DC/frml-154

Adds scripts to spin up local replica Label Studio Server

src/frdc/load/label_studio.py

@@ -16,39 +16,58 @@ def get_bounds_and_labels(self) -> tuple[list[tuple[int, int]], list[str]]:
         bounds = []
         labels = []
 
-        ann = self["annotations"][0]
-        results = ann["result"]
-        for r_ix, r in enumerate(results):
-            r: dict
+        # Each annotation is an entire image labelled by a single person.
+        # By selecting the 0th annotation, we are usually selecting the main
+        # annotation.
+        annotation = self["annotations"][0]
+
+        # There are some metadata in `annotation`, but we just want the results
+        results = annotation["result"]
+
+        for bbox_ix, bbox in enumerate(results):
+            # 'id' = {str} 'jr4EXAKAV8'
+            # 'type' = {str} 'polygonlabels'
+            # 'value' = {dict: 3} {
+            #       'closed': True,
+            #       'points': [[x0, y0], [x1, y1], ... [xn, yn]],
+            #       'polygonlabels': ['label']
+            # }
+            # 'origin' = {str} 'manual'
+            # 'to_name' = {str} 'image'
+            # 'from_name' = {str} 'label'
+            # 'image_rotation' = {int} 0
+            # 'original_width' = {int} 450
+            # 'original_height' = {int} 600
+            bbox: dict
 
             # See Issue FRML-78: Somehow some labels are actually just metadata
-            if r["from_name"] != "label":
+            if bbox["from_name"] != "label":
                 continue
 
             # We flatten the value dict into the result dict
-            v = r.pop("value")
-            r = {**r, **v}
+            v = bbox.pop("value")
+            bbox = {**bbox, **v}
 
             # Points are in percentage, we need to convert them to pixels
-            r["points"] = [
+            bbox["points"] = [
                 (
-                    int(x * r["original_width"] / 100),
-                    int(y * r["original_height"] / 100),
+                    int(x * bbox["original_width"] / 100),
+                    int(y * bbox["original_height"] / 100),
                 )
-                for x, y in r["points"]
+                for x, y in bbox["points"]
             ]
 
             # Only take the first label as this is not a multi-label task
-            r["label"] = r.pop("polygonlabels")[0]
-            if not r["closed"]:
+            bbox["label"] = bbox.pop("polygonlabels")[0]
+            if not bbox["closed"]:
                 logger.warning(
-                    f"Label for {r['label']} @ {r['points']} not closed. "
+                    f"Label for {bbox['label']} @ {bbox['points']} not closed. "
                     f"Skipping"
                 )
                 continue
 
-            bounds.append(r["points"])
-            labels.append(r["label"])
+            bounds.append(bbox["points"])
+            labels.append(bbox["label"])
 
         return bounds, labels
 

src/label-studio/label-studio-replica/default_config.xml

@@ -0,0 +1,121 @@
+<View>
+    <Header size="2">Replica FRDC Server</Header>
+  	<Header size="6" style="color:red">
+      This is a replica server. All changes will NOT be reflected onto the Machine Learning Pipeline.
+  	</Header>
+    <Image name="image" value="$img" zoom="true" zoomControl="true"/>
+    <Header size="8">Select Species</Header>
+    <View style="height:200px; overflow:auto;">
+        <Filter name="filterSpecies" toName="label" hotkey="shift+f"
+                minlength="0" placeholder="Type Species Name"/>
+        <PolygonLabels name="label" toName="image" strokeWidth="3"
+                       pointSize="small" opacity="0.1" required="True">
+
+            <Label value="Alstonia Angulstiloba" background="#FFA39E"/>
+            <Label value="Bridelia Sp." background="#D4380D"/>
+            <Label value="Calophyllum" background="#FFC069"/>
+            <Label value="Campnosperma Auriculatum" background="#AD8B00"/>
+            <Label value="Cinnamomum Iners" background="#D3F261"/>
+            <Label value="Claoxylon Indicum" background="#389E0D"/>
+            <Label value="Clausena Excavata" background="#5CDBD3"/>
+            <Label value="Cratoxylum Formosum" background="#096DD9"/>
+            <Label value="Dillenia Suffruticosa" background="#ADC6FF"/>
+            <Label value="Fagraea fragans" background="#389E0D"/>
+            <Label value="Falcataria Moluccana" background="#9254DE"/>
+            <Label value="Ficus Variegata" background="#F759AB"/>
+            <Label value="Ixonanthes reticulata" background="#008BAD"/>
+            <Label value="Koompassia malaccensis" background="#FFA39E"/>
+            <Label value="Leea Indica" background="#FFA39E"/>
+            <Label value="Litsea firma" background="#F8FE00"/>
+            <Label value="Macaranga Gigantea" background="#D4380D"/>
+            <Label value="Palm Tree" background="#FFC069"/>
+            <Label value="Pennisetum Purpureum" background="#AD8B00"/>
+            <Label value="Pometia Pinnata" background="#D3F261"/>
+            <Label value="Prunus polystachya" background="#F3A201"/>
+            <Label value="Sandoricum Koetjape" background="#389E0D"/>
+            <Label value="Shorea Leprosula" background="#5CDBD3"/>
+            <Label value="Spathodea Campanulatum" background="#096DD9"/>
+            <Label value="Sterculia Parviflora" background="#E178C5"/>
+            <Label value="Syzygium grande" background="#5C0BD3"/>
+            <Label value="Syzygium Polyanthum" background="#9254DE"/>
+            <Label value="Terminalia Catappa" background="#F759AB"/>
+            <Label value="_Unknown" background="#676767"/>
+        </PolygonLabels>
+    </View>
+
+    <View visibleWhen="region-selected">
+        <Header>Select Quality</Header>
+        <Choices name="Quality" toName="image" perRegion="true"
+                 choice="single-radio" required="false">
+            <Choice value="Good" background="green"/>
+            <Choice value="Bad" background="red"/>
+        </Choices>
+
+        <Choices name="BadRatingReason" toName="image" choice="multiple"
+                 showInLine="true" visibleWhen="choice-selected"
+                 whenTagName="Quality" whenChoiceValue="Bad" perRegion="true">
+
+            <Choice value="Artefacts Present"/>
+            <Choice value="Brightness Problem"/>
+            <Choice value="Others (write in comment)"/>
+
+            <View visibleWhen="choice-selected" whenTagName="BadRatingReason"
+                  whenChoiceValue="Others (write in comment)">
+                <TextArea name="BadRatingReasonComment" toName="image"
+                          perRegion="true" placeholder="Write the reason">
+                </TextArea>
+            </View>
+
+        </Choices>
+        <Header>Submitted By (Team):</Header>
+        <View>
+            <!--Filter name="SubmitTeamFilter" toName="Team" hotkey="shift+f" minlength="0" placeholder="type team name"/ -->
+            <Labels required="false" name="Team" toName="image"
+                    showInline="false" perRegion="true">
+                <Label value="EngineeringSubmit"/>
+                <Label value="EcologySubmit"/>
+            </Labels>
+            <Header size="7">UserID (Submit):</Header>
+            <!--TextArea required="false" name="SubmitUserId" toName="image" placeholder="userID" perRegion="true"/ -->
+            <Choices name="SubmitUserIdChoice" toName="Team" perRegion="true"
+                     choice="single-radio" visibleWhen="region-selected">
+                <Choice value="jjsit"/>
+                <Choice value="shyap"/>
+                <Choice value="jchang"/>
+                <Choice value="bchew"/>
+                <Choice value="yksim"/>
+                <Choice value="ahassan"/>
+            </Choices>
+        </View>
+
+        <Header>Checked By (Team):</Header>
+        <View>
+            <!--Filter name="CheckTeamFilter" toName="CheckTeam" hotkey="shift+f" minlength="0" placeholder="type team name"/ -->
+            <Labels required="false" name="CheckTeam" toName="image"
+                    showInline="false" perRegion="true">
+                <Label value="EngineeringChecked_Agree"/>
+                <Label value="EcologyChecked_Agree"/>
+                <Label value="EngineeringChecked_Disagree"/>
+                <Label value="EcologyChecked_Disagree"/>
+
+
+            </Labels>
+            <Header size="7">UserID (Check):</Header>
+            <TextArea name="CheckUserId" toName="image" placeholder="userID"
+                      perRegion="true"/>
+        </View>
+        <Header>General Comment</Header>
+        <View>
+            <Labels required="false" name="CommentType" toName="image"
+                    showInline="false" perRegion="true">
+                <Label value="New Comment"/>
+                <Label value="Resolved Comment"/>
+            </Labels>
+            <TextArea name="GeneralComment" toName="image" perRegion="true"
+                      placeholder="General Comment">
+            </TextArea>
+        </View>
+
+    </View>
+
+</View>

src/label-studio/label-studio-replica/docker-compose.yml

@@ -0,0 +1,78 @@
+version: "3.9"
+services:
+  nginx:
+    build: .
+    image: heartexlabs/label-studio:latest
+    restart: unless-stopped
+    ports:
+      - "8082:8085"
+      - "8083:8086"
+    depends_on:
+      - app
+    environment:
+      - LABEL_STUDIO_HOST=${LABEL_STUDIO_HOST:-}
+    #   Optional: Specify SSL termination certificate & key
+    #   Just drop your cert.pem and cert.key into folder 'deploy/nginx/certs'
+    #      - NGINX_SSL_CERT=/certs/cert.pem
+    #      - NGINX_SSL_CERT_KEY=/certs/cert.key
+    volumes:
+      - ./mydata:/label-studio/data:rw
+      - ./deploy/nginx/certs:/certs:ro
+    #   Optional: Override nginx default conf
+    #      - ./deploy/my.conf:/etc/nginx/nginx.conf
+    command: nginx
+    networks:
+      - label-studio-dev
+
+  app:
+    stdin_open: true
+    tty: true
+    build: .
+    image: heartexlabs/label-studio:latest
+    restart: unless-stopped
+    expose:
+      - "8000"
+    depends_on:
+      - db-dev
+    environment:
+      - DJANGO_DB=default
+      - POSTGRE_NAME=postgres
+      - POSTGRE_USER=postgres
+      - POSTGRE_PASSWORD=
+      - POSTGRE_PORT=5432
+      - POSTGRE_HOST=db-dev
+      - LABEL_STUDIO_HOST=${LABEL_STUDIO_HOST:-}
+      - JSON_LOG=1
+    #      - LOG_LEVEL=DEBUG
+    volumes:
+      - ./mydata:/label-studio/data:rw
+    command: label-studio-uwsgi
+    networks:
+      - label-studio-dev
+
+  db-dev:
+    image: postgres:11.5
+    hostname: db-dev
+    restart: unless-stopped
+    # Optional: Enable TLS on PostgreSQL
+    # Just drop your server.crt and server.key into folder 'deploy/pgsql/certs'
+    # NOTE: Both files must have permissions u=rw (0600) or less
+    #    command: >
+    #      -c ssl=on
+    #      -c ssl_cert_file=/var/lib/postgresql/certs/server.crt
+    #      -c ssl_key_file=/var/lib/postgresql/certs/server.key
+    ports:
+      - "5435:5432"
+    environment:
+      - POSTGRES_HOST_AUTH_METHOD=trust
+    volumes:
+      - ${POSTGRES_DATA_DIR:-./postgres-data}:/var/lib/postgresql/data
+      - ${POSTGRES_DATA_DIR:-./postgres-backups}:/var/lib/postgresql/backups
+      - ./deploy/pgsql/certs:/var/lib/postgresql/certs:ro
+    networks:
+      - label-studio-dev
+
+networks:
+  label-studio-dev:
+    name: label-studio-dev
+    driver: bridge

src/label-studio/label-studio-replica/initialize_replica.py

@@ -0,0 +1,99 @@
+import os
+import time
+from pathlib import Path
+
+import label_studio_sdk
+
+THIS_DIR = Path(__file__).parent
+
+# This is your API token. I put mine here, which is OK only if you're in a
+# development environment. Otherwise, do not.
+dev_api_key = os.getenv("REPLICA_LABEL_STUDIO_API_KEY")
+prd_api_key = os.getenv("LABEL_STUDIO_API_KEY")
+dev_url = "http://localhost:8082"
+prd_url = "http://localhost:8080"
+
+# We can initialize the sdk using this following.
+# The client is like the middleman between you as a programmer, and the
+# Label Studio (LS) server.
+dev_client = label_studio_sdk.Client(url=dev_url, api_key=dev_api_key)
+prd_client = label_studio_sdk.Client(url=prd_url, api_key=prd_api_key)
+
+# This is the labelling interface configuration.
+# We can save it somewhere as an XML file then import it too
+dev_config = (THIS_DIR / "default_config.xml").read_text()
+
+# %%
+print("Creating Development Project...")
+# Creates the project, note to set the config here
+dev_proj = dev_client.create_project(
+    title="FRDC Replica",
+    description="This is the replica project of FRDC. It's ok to break this.",
+    label_config=dev_config,
+    color="#FF0025",
+)
+# %%
+print("Adding Import Source...")
+# This links to our GCS as an import source
+dev_storage = dev_proj.connect_google_import_storage(
+    bucket="frdc-ds",
+    regex_filter=".*.jpg",
+    google_application_credentials=(
+        THIS_DIR / "frmodel-943e4feae446.json"
+    ).read_text(),
+    presign=False,
+    title="Source",
+)
+time.sleep(5)
+# %%
+print("Syncing Storage...")
+# Then, we sync it so that all the images appear as annotation targets
+dev_proj.sync_storage(
+    storage_type=dev_storage["type"],
+    storage_id=dev_storage["id"],
+)
+time.sleep(5)
+# %%
+print("Retrieving Tasks...")
+prd_proj = prd_client.get_project(id=1)
+prd_tasks = prd_proj.get_tasks()
+dev_tasks = dev_proj.get_tasks()
+# %%
+# This step copies over the annotations from the production to the development
+# This creates it as a "prediction"
+print("Copying Annotations...")
+for prd_task in prd_tasks:
+    # For each prod task, we find the corresponding (image) file name
+    prd_fn = prd_task["storage_filename"]
+
+    # Then, we find the corresponding task in the development project
+    dev_tasks_matched = [
+        t for t in dev_tasks if t["storage_filename"] == prd_fn
+    ]
+
+    # Do some error handling
+    if len(dev_tasks_matched) == 0:
+        print(f"File not found in dev: {prd_fn}")
+        continue
+    if len(dev_tasks_matched) > 1:
+        print(f"Too many matches found in dev: {prd_fn}")
+        continue
+
+    # Get the first match
+    dev_task = dev_tasks_matched[0]
+
+    # Only get annotations by evening
+    prd_ann = [
+        ann
+        for ann in prd_task["annotations"]
+        if "dev_evening" in ann["created_username"]
+    ][0]
+
+    # Create the prediction using the result from production
+    dev_proj.create_prediction(
+        task_id=dev_task["id"],
+        result=prd_ann["result"],
+        model_version="API Testing Prediction",
+    )
+
+print("Done!")