Prediction Feature

.gitignore

@@ -140,3 +140,6 @@ GitHub.sublime-settings
 
 # certbot configuration
 certbot/*
+
+# gcloud credential
+credential*.json

app/Dockerfile

@@ -13,6 +13,8 @@ RUN pip install --upgrade pip
 COPY ./requirements.txt .
 RUN pip install -r requirements.txt
 
+RUN set TF_ENABLE_ONEDNN_OPTS=0
+
 # copy entrypoint.sh
 COPY ./entrypoint.sh .
 RUN sed -i 's/\r$//g' /usr/src/app/entrypoint.sh

app/ce_tour/gcloud.py

@@ -0,0 +1,19 @@
+from urllib.parse import urljoin
+
+from django.conf import settings
+from storages.backends.gcloud import GoogleCloudStorage
+from storages.utils import setting
+
+
+class GoogleCloudMediaFileStorage(GoogleCloudStorage):
+    """
+    Google file storage class which gives a media file path from       MEDIA_URL not google generated one.
+    """
+
+    bucket_name = setting("GS_BUCKET_NAME")
+
+    def url(self, name):
+        """
+        Gives correct MEDIA_URL and not google generated url.
+        """
+        return urljoin(settings.MEDIA_URL, name)

app/ce_tour/settings.py

@@ -111,6 +111,17 @@
     },
 ]
 
+# configuration for gcloud storage
+from google.oauth2 import service_account
+
+GS_CREDENTIALS = service_account.Credentials.from_service_account_file(
+    os.getenv("GS_CREDENTIALS")
+)
+
+DEFAULT_FILE_STORAGE = "ce_tour.gcloud.GoogleCloudMediaFileStorage"
+GS_PROJECT_ID = os.getenv("GS_PROJECT_ID")
+GS_BUCKET_NAME = os.getenv("GS_BUCKET_NAME")
+MEDIA_URL = "https://storage.googleapis.com/{}/".format(GS_BUCKET_NAME)
 
 # Internationalization
 # https://docs.djangoproject.com/en/5.0/topics/i18n/

app/prediction/__init__.py

@@ -0,0 +1,63 @@
+import logging
+import tempfile
+
+from django.conf import settings
+from google.cloud import storage
+
+from .prediction import Prediction
+
+LABELS = sorted(
+    [
+        "benteng_vredeburg",
+        "candi_borobudur",
+        "candi_prambanan",
+        "garuda_wisnu_kencana",
+        "gedung_sate",
+        "istana_maimun",
+        "jam_gadang",
+        "keong_mas",
+        "keraton_jogja",
+        "kota_tua",
+        "lawang_sewu",
+        "masjid_istiqlal",
+        "masjid_menara_kudus",
+        "masjid_raya_baiturrahman",
+        "menara_siger_lampung",
+        "monas",
+        "monumen_bandung_lautan_api",
+        "monumen_gong_perdamaian",
+        "monumen_nol_km",
+        "monumen_simpang_lima_gumul",
+        "patung_ikan_surabaya",
+        "patung_yesus_memberkati",
+        "tugu_jogja",
+        "tugu_khatulistiwa",
+        "tugu_pahlawan_surabaya",
+    ]
+)
+
+# # Initialize the model
+# Model = Prediction(
+#     model_path="./prediction/model_predic/best_model_2.h5",
+#     target_size=(224, 224),
+#     classes=LABELS)
+
+try:
+    client = storage.Client(
+        credentials=settings.GS_CREDENTIALS, project=settings.GS_PROJECT_ID
+    )
+    bucket = client.get_bucket(settings.GS_BUCKET_NAME)
+    # Construct the source path in GCS
+    gcs_source_path = "model/best_model_2.h5"
+
+    # Download the model file from GCS into a temporary file
+    with tempfile.NamedTemporaryFile(suffix=".h5") as temp_model_file:
+        blob = bucket.blob(gcs_source_path)
+        blob.download_to_filename(temp_model_file.name)
+
+        # Use the downloaded temporary file as the model path
+        Model = Prediction(
+            model_path=temp_model_file.name, target_size=(224, 224), classes=LABELS
+        )
+except Exception as e:
+    logger.error(f"Error reading from GCS: {e}")

app/prediction/migrations/0001_initial.py

@@ -0,0 +1,25 @@
+# Generated by Django 5.0.6 on 2024-06-02 18:27
+
+import prediction.models
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    initial = True
+
+    dependencies = [
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='Prediction',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('result', models.CharField(max_length=200)),
+                ('rate', models.DecimalField(decimal_places=2, max_digits=4)),
+                ('image', models.URLField()),
+                ('createdAt', models.DateTimeField(auto_now_add=True)),
+            ],
+        ),
+    ]

app/prediction/models.py

@@ -1,3 +1,8 @@
 from django.db import models
 
-# Create your models here.
+
+class Prediction(models.Model):
+    result = models.CharField(max_length=200)
+    rate = models.DecimalField(max_digits=4, decimal_places=2)
+    image = models.URLField()
+    createdAt = models.DateTimeField(auto_now_add=True)

app/prediction/prediction.py

@@ -0,0 +1,33 @@
+import tensorflow as tf
+
+
+class Prediction:
+    def __init__(self, model_path: str, classes: list, target_size: tuple = (224, 224)):
+        self.model_path = model_path
+        self.classes = classes
+        self.target_size = target_size
+
+        try:
+            self.__loaded_model = tf.keras.models.load_model(self.model_path)
+        except Exception as e:
+            print(e)
+
+        self.__prepared_img = None
+
+    def model_summary(self):
+        return self.__loaded_model.summary()
+
+    def __preprocess_img(self, img):
+        img = tf.io.read_file(img)
+        img = tf.io.decode_image(img)
+        img = tf.image.resize(img, self.target_size)
+        return img
+
+    def predict_class(self, img):
+        img = self.__preprocess_img(img)
+        pred_prob = self.__loaded_model.predict(tf.expand_dims(img, axis=0))
+        pred_cat = pred_prob.argmax(axis=-1)
+        pred_class = self.classes[pred_cat[0]]
+        confidence_score = pred_prob.max() * 100
+        confidence_score = f"{confidence_score:.2f}"
+        return pred_class, confidence_score

app/prediction/serializers.py

@@ -0,0 +1,8 @@
+from prediction.models import Prediction
+from rest_framework import serializers
+
+
+class PredictionSerializer(serializers.ModelSerializer):
+    class Meta:
+        model = Prediction
+        fields = "__all__"

app/prediction/urls.py

@@ -1 +1,8 @@
 urlpatterns = []
+from django.urls import path
+
+from .views import AddCapture
+
+urlpatterns = [
+    path("", AddCapture.as_view(), name="landmark"),
+]

app/prediction/views.py

@@ -1,3 +1,89 @@
-from django.shortcuts import render
+import logging
+import os
+import uuid
 
-# Create your views here.
+from django.conf import settings
+from google.cloud import storage
+from rest_framework import status
+from rest_framework.response import Response
+from rest_framework.views import APIView
+
+from . import Model
+from .serializers import PredictionSerializer
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+
+class AddCapture(APIView):
+    def post(self, request):
+        if "image" not in request.FILES:
+            return Response(
+                {"message": "No file provided"}, status=status.HTTP_400_BAD_REQUEST
+            )
+
+        save_directory = "/photo/"
+
+        if not os.path.exists(save_directory):
+            os.makedirs(save_directory)
+
+        uploaded_file = request.FILES["image"]
+
+        new_filename = f"{uuid.uuid4()}.{uploaded_file.name.split('.')[-1]}"
+
+        file_path = os.path.join(save_directory, new_filename)
+
+        with open(file_path, "wb") as f:
+            for chunk in uploaded_file.chunks():
+                f.write(chunk)
+
+        try:
+            predicted_class_name, confidence_score = Model.predict_class(file_path)
+
+            if float(confidence_score) < 65.0:
+                return Response(
+                    {
+                        "message": "Confidence score below 65%. Please provide a clearer image."
+                    },
+                    status=status.HTTP_400_BAD_REQUEST,
+                )
+        except Exception as e:
+            logger.error(f"Error in prediction: {e}")
+            return Response(
+                {"message": str(e)}, status=status.HTTP_500_INTERNAL_SERVER_ERROR
+            )
+
+        try:
+            client = storage.Client(
+                credentials=settings.GS_CREDENTIALS, project=settings.GS_PROJECT_ID
+            )
+            bucket = client.get_bucket(settings.GS_BUCKET_NAME)
+            gcs_destination_path = f"predictions/{predicted_class_name}/{new_filename}"
+
+            blob = bucket.blob(gcs_destination_path)
+            blob.upload_from_filename(file_path)
+            image_url = blob.public_url
+        except Exception as e:
+            logger.error(f"Error uploading to GCS: {e}")
+            return Response(
+                {"message": str(e)}, status=status.HTTP_500_INTERNAL_SERVER_ERROR
+            )
+
+        os.remove(file_path)
+
+        if os.path.exists(file_path) == 0:
+            print("File does not exist")
+
+        data = {
+            "result": predicted_class_name,
+            "rate": confidence_score,
+            "image": image_url,
+        }
+        photo_serializer = PredictionSerializer(data=data)
+
+        if photo_serializer.is_valid():
+            photo_serializer.save()
+            return Response(photo_serializer.data, status=status.HTTP_201_CREATED)
+        else:
+            logger.error(f"Serializer errors: {photo_serializer.errors}")
+            return Response(photo_serializer.errors, status=status.HTTP_400_BAD_REQUEST)