Feat yolo classif

backend/src/main.py

@@ -4,6 +4,7 @@
 import time
 from datetime import datetime
 from logging.handlers import TimedRotatingFileHandler
+from contextlib import asynccontextmanager
 from typing import Union
 from uuid import uuid4
 
@@ -159,12 +160,7 @@ async def add_owasp_middleware(request: Request, call_next):
 logger = setup_logs(PATH_LOGS)
 
 # Load model
-MODEL_PATH = os.path.join(CURRENT_DIR, "../model.pt")
-model = None
-if os.path.exists(MODEL_PATH):
-    model = load_model_inference(MODEL_PATH)
-if not model:
-    raise RuntimeError("Model not found")
+app.model = load_model_inference("./model.pt")
 
 # Object storage
 S3_URL_ENDPOINT = os.environ["S3_URL_ENDPOINT"]
@@ -173,13 +169,6 @@ async def add_owasp_middleware(request: Request, call_next):
 
 s3 = boto3.resource("s3", endpoint_url=S3_URL_ENDPOINT, verify=False)
 
-""" TODO : check if connection successful
-try:
-    s3.meta.client.head_bucket(Bucket=S3_BUCKET_NAME)
-except ClientError:
-    logger.exception("Cannot find s3 bucket ! Are you sure your credentials are correct ?")
-"""
-
 # Versions
 if "versions.json" in os.listdir(os.path.dirname(CURRENT_DIR)):
     with open("versions.json", "r") as f:
@@ -238,13 +227,13 @@ async def imageupload(
 
         # send image to model for prediction
         start = time.time()
-        label, confidence = predict_image(model, img_bytes)
+        label, confidence = predict_image(app.model, img_bytes)
         extras_logging["bg_label"] = label
         extras_logging["bg_confidence"] = confidence
         extras_logging["bg_model_time"] = round(time.time() - start, 2)
         if confidence < 0.76:
             extras_logging["bg_confidence_level"] = "low"
-        elif confidence < 0.99:
+        elif confidence < 0.98:
             extras_logging["bg_confidence_level"] = "medium"
         else:
             extras_logging["bg_confidence_level"] = "high"

backend/src/model.py

@@ -28,10 +28,9 @@ def load_model_inference(model_path: str):
         model_path (str): path to model (.pt file)
 
     Returns:
-        Model: loaded model ready for prediction
+        Model: loaded model ready for prediction and Warm-up
     """
-    model = YOLO(model_path)
-    return model
+    return YOLO(model_path)
 
 
 def predict_image(model, img: bytes) -> Union[str, float]:

backend/tests/test_model.py

@@ -2,17 +2,12 @@
 
 import pytest
 from src.model import CLASSES, load_model_inference, predict_image
-
+from src.main import app
 
 class TestModel:
-    model_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../model.pt")
-    assert os.path.exists(model_path)
-    model = load_model_inference(model_path)
-
     def test_predict_image(self):
         """Checks the prediction of an image by the model"""
-        path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "revolver.jpg")
-        with open(path, "rb") as f:
-            res = predict_image(self.model, f.read())
+        with open("./tests/revolver.jpg", "rb") as f:
+            res = predict_image(app.model, f.read())
         assert res[0] == "revolver"
         assert res[1] == pytest.approx(1, 0.1)

frontend/src/components/ResultPage.vue

@@ -129,14 +129,14 @@ function sendFeedback (isCorrect: boolean) {
                 <DsfrTag
                   class="fr-tag--sm success-tag"
                 >
-                  Indice de fiabilité : {{ Math.floor(confidence) }}%
+                  Indice de fiabilité : {{ Math.floor(confidence * 100) }}%
                 </DsfrTag>
               </div>
               <div v-else>
                 <DsfrTag
                   class="fr-tag--sm warning-tag"
                 >
-                  Indice de fiabilité : {{ Math.floor(confidence) }}%
+                  Indice de fiabilité : {{ Math.floor(confidence * 100) }}%
                 </DsfrTag>
                 <p class="warning-text">
                   Nous vous conseillons de faire appel à un expert pour confirmer cette réponse.