Revert "feat: use the new category classifier model for category pred…

…iction (#512)"

This reverts commit a45ded0.

.github/workflows/models-deploy.yml

@@ -75,3 +75,34 @@ jobs:
             tar -xzvf saved_model.tar.gz --strip-component=1
             rm saved_model.tar.gz
 
+  # TODO(kulizhsy): delete this job after we've migrated to TF Serving.
+  keras-category-classifier-sync:
+    name: Sync Keras category classifier configs
+    runs-on: ubuntu-latest
+    environment: ${{ matrix.env }}
+    strategy:
+      matrix:
+        env:
+          - ${{ startsWith(github.ref, 'refs/tags/v') && 'robotoff-org' || 'robotoff-net' }}
+    steps:
+      - uses: actions/checkout@master
+      - name: Download
+        uses: appleboy/ssh-action@master
+        with:
+          host: ${{ secrets.HOST }}
+          username: ${{ secrets.USERNAME }}
+          key: ${{ secrets.SSH_PRIVATE_KEY }}
+          proxy_host: ${{ secrets.PROXY_HOST }}
+          proxy_username: ${{ secrets.USERNAME }}
+          proxy_key: ${{ secrets.SSH_PRIVATE_KEY }}
+          script: |
+            cd ${{ matrix.env }}/models
+            base_url=https://github.com/openfoodfacts/robotoff-models/releases/download/keras-category-classifier-xx-1.0
+            dest_folder=category
+            mkdir -p ${dest_folder}
+            wget -cO - ${base_url}/config.json > ${dest_folder}/config.json
+            wget -cO - ${base_url}/category_taxonomy.json > ${dest_folder}/category_taxonomy.json
+            wget -cO - ${base_url}/category_voc.json > ${dest_folder}/category_voc.json
+            wget -cO - ${base_url}/ingredient_voc.json > ${dest_folder}/ingredient_voc.json
+            wget -cO - ${base_url}/product_name_voc.json > ${dest_folder}/product_name_voc.json
+            wget -cO - ${base_url}/checkpoint.hdf5 > ${dest_folder}/checkpoint.hdf5

data/clf_category_blacklist.txt

@@ -0,0 +1,26 @@
+en:alcoholic-beverages
+en:beverages
+en:biscuits-and-cakes
+en:breakfasts
+en:cereals-and-potatoes
+en:cereals-and-their-products
+en:dairies
+en:fermented-foods
+en:fermented-milk-products
+en:frozen-foods
+en:fruits-and-vegetables-based-foods
+en:fruits-based-foods
+en:groceries
+en:meals
+en:meat-based-products
+en:meats
+en:non-alcoholic-beverages
+en:plant-based-beverages
+en:plant-based-foods
+en:plant-based-foods-and-beverages
+en:prepared-meats
+en:snacks
+en:spreads
+en:sweet-snacks
+en:sweet-spreads
+en:vegetables-based-foods

robotoff/cli/main.py

@@ -194,31 +194,25 @@ def download_models(force: bool = False) -> None:
 def categorize(
     barcode: str,
     deepest_only: bool = False,
+    blacklist: bool = False,
 ) -> None:
-    """Categorise predicts product categories based on the neural category classifier.
-
-    deepest_only: controls whether the returned predictions should only contain the deepmost
-    categories for a predicted taxonomy chain.
-    For example, if we predict 'fresh vegetables' -> 'legumes' -> 'beans' for a product,
-    setting deepest_only=True will return 'beans'."""
-    from robotoff.ml.category.neural.category_classifier import CategoryClassifier
-    from robotoff.products import get_product
-    from robotoff.taxonomy import TaxonomyType, get_taxonomy
-
-    product = get_product(barcode)
-    if product is None:
-        print(f"Product {barcode} not found")
-        return
-
-    predicted = CategoryClassifier(get_taxonomy(TaxonomyType.category.name)).predict(
-        product, deepest_only
+    from robotoff import settings
+    from robotoff.ml.category.neural.model import (
+        LocalModel,
+        filter_blacklisted_categories,
     )
+    from robotoff.utils import get_logger
+
+    get_logger()
+    model = LocalModel(settings.CATEGORY_CLF_MODEL_PATH)
+    predicted = model.predict_from_barcode(barcode, deepest_only=deepest_only)
 
     if predicted:
-        for prediction in predicted:
-            print(f"{prediction.category}: {prediction.confidence}")
-    else:
-        print(f"Nothing predicted for product {barcode}")
+        if blacklist:
+            predicted = filter_blacklisted_categories(predicted)
+
+        for cat, confidence in predicted:
+            print("{}: {}".format(cat, confidence))
 
 
 @app.command()

robotoff/insights/importer.py

@@ -73,7 +73,6 @@ def import_insights(
         server_domain: str,
         automatic: bool,
     ) -> int:
-        """Returns the number of insights that were imported."""
         timestamp = datetime.datetime.utcnow()
         processed_insights: Iterator[Insight] = self.process_insights(
             data, server_domain, automatic