Update app.py

app.py

@@ -1,74 +1,31 @@
-import os
-import random
+from fastapi import FastAPI, UploadFile, File
+from fastapi.responses import JSONResponse
+from PIL import Image
 import numpy as np
 import tensorflow as tf
-import torch
-
-# Ensure deterministic execution and avoid random_device
-os.environ["PYTHONHASHSEED"] = "0"
-random.seed(0)
-np.random.seed(0)
-tf.random.set_seed(0)
-torch.manual_seed(0)
-torch.use_deterministic_algorithms(True)
-torch.set_deterministic_debug_mode("warn")
-
-
-from segment_anything import sam_model_registry, SamPredictor
+import io
 
 app = FastAPI()
 
-# Your secret API key (you can also load this from .env or environment variables)
-API_KEY = "my-secret-api-key"
-
-# Load the classification model
-model = tf.keras.models.load_model("1.keras")
+# Load the model
+model = tf.keras.models.load_model("2.keras")
 CLASS_NAMES = ['Fungi', 'Healthy', 'Nematode', 'Pest', 'Phytopthora', 'Virus']
 
-# Load the SAM model
-DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
-SAM_CHECKPOINT_PATH = "C:/Users/lenovo/Downloads/models/sam_model/sam_vit_b.pth"
-
-sam = sam_model_registry["vit_b"](checkpoint=SAM_CHECKPOINT_PATH).to(DEVICE)
-predictor = SamPredictor(sam)
-
 @app.post("/predict")
-async def predict(
-    file: UploadFile = File(...),
-    x_api_key: str = Header(None)
-):
-    # Check API key
-    if x_api_key != API_KEY:
-        raise HTTPException(status_code=401, detail="Invalid API Key")
-
+async def predict(file: UploadFile = File(...)):
     try:
-        # Load and prepare the input image
         contents = await file.read()
-        image = Image.open(io.BytesIO(contents)).convert("RGB")
-        image_np = np.array(image)
-
-        # Convert to BGR for SAM
-        image_bgr = cv2.cvtColor(image_np, cv2.COLOR_RGB2BGR)
 
-        # Run SAM segmentation
-        predictor.set_image(image_bgr)
-        H, W, _ = image_bgr.shape
-        input_point = np.array([[W // 2, H // 2]])
-        input_label = np.array([1])
-        masks, scores, _ = predictor.predict(point_coords=input_point, point_labels=input_label, multimask_output=False)
-
-        # Apply the mask
-        mask = masks[0]
-        segmented = image_np * np.expand_dims(mask, axis=-1)
+        # Load and process image
+        image = Image.open(io.BytesIO(contents)).convert("RGB")
+        image = image.resize((224, 224))
 
-        # Prepare for classification
-        segmented_pil = Image.fromarray(segmented.astype("uint8")).convert("RGB")
-        segmented_pil = segmented_pil.resize((224, 224))
-        arr = np.array(segmented_pil).astype("float32")
-        arr = np.expand_dims(arr, axis=0)
+        # Don't normalize — match tf.data image_dataset_from_directory
+        img_array = np.array(image).astype("float32")
+        img_array = np.expand_dims(img_array, axis=0)  # (1, 224, 224, 3)
 
         # Predict
-        prediction = model.predict(arr)
+        prediction = model.predict(img_array)
         predicted_class = int(np.argmax(prediction[0]))
         predicted_label = CLASS_NAMES[predicted_class]