Fix dictionary mapped strings for Specialist classes

backend/app.py

@@ -193,18 +193,16 @@ SPECIALIST_MAP = {
     "Corn (Maize) with Cercospora and Gray Leaf Spot": "Corn - Gray Leaf Spot",
     "Corn (Maize) with Common Rust": "Corn - Common Rust",
     "Corn (Maize) with Northern Leaf Blight": "Corn - Northern Leaf Blight",
-    "Healthy Corn": "Corn - Healthy",
-    "Grape with Black Measles": "Grape - Black Measles",
+    "Healthy Corn (Maize) Plant": "Corn - Healthy",
     "Grape with Black Rot": "Grape - Black Rot",
-    "Grape with Leaf Blight": "Grape - Leaf Blight",
-    "Healthy Grape": "Grape - Healthy",
+    "Grape with Esca (Black Measles)": "Grape - Black Measles",
+    "Grape with Isariopsis Leaf Spot": "Grape - Leaf Blight",
+    "Healthy Grape Plant": "Grape - Healthy",
     "Orange with Citrus Greening": "Orange - Citrus Greening",
     "Peach with Bacterial Spot": "Peach - Bacterial Spot",
-    "Healthy Peach": "Peach - Healthy",
-    "Pepper Bell with Bacterial Spot": "Pepper - Bacterial Spot",
+    "Healthy Peach Plant": "Peach - Healthy",
     "Bell Pepper with Bacterial Spot": "Pepper - Bacterial Spot",
-    "Healthy Pepper Bell": "Pepper - Healthy",
-    "Healthy Bell Pepper": "Pepper - Healthy",
+    "Healthy Bell Pepper Plant": "Pepper - Healthy",
     "Potato with Early Blight": "Potato - Early Blight",
     "Potato with Late Blight": "Potato - Late Blight",
     "Healthy Potato Plant": "Potato - Healthy",

test_hf_automated.py

@@ -0,0 +1,82 @@
+import requests
+import io
+import time
+import urllib.request
+import json
+from PIL import Image
+
+API_URL = "https://rhamprassath-greendoctor-backend.hf.space/predict"
+
+# Known URLs of raw PlantVillage images from a public repo
+TEST_IMAGES = {
+    # Specialist Examples
+    "Tomato - Septoria": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Tomato___Septoria_leaf_spot/0a555f63-bf03-4958-8993-e1932b8dce9f___JR_Sept.L.S%208395.JPG",
+    "Tomato - Early Blight": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Tomato___Early_blight/0a0a5247-c035-42a9-aa5a-b6e828453ad2___RS_Erly.B%209425.JPG",
+    "Potato - Late Blight": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Potato___Late_blight/0a1a084c-ddbe-4767-a06f-13589be76228___RS_LB%202758.JPG",
+    "Apple - Scab": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Apple___Apple_scab/0a5e9323-dbad-432d-ac58-d291718345d9___FREC_Scab%203417.JPG",
+    "Corn - Common Rust": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Corn_(maize)___Common_rust_/RS_Rust%201676.JPG",
+    "Grape - Black Rot": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Grape___Black_rot/0a283e58-f9b6-4993-bc89-10659345712c___FAM_B.Rot%200585.JPG",
+    "Peach - Bacterial Spot": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Peach___Bacterial_spot/0a158b43-2df7-463d-bacc-39bf7238a204___Rutg._Bact.S%201402.JPG",
+    
+    # Healthy Varieties
+    "Tomato - Healthy": "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Tomato___healthy/0a0bb533-5c74-4286-90ae-2fba6b364429___GH_Hlthy%209689.JPG",
+    
+    # We also have Generalist classes like Rice and Wheat, but those aren't in PlantVillage exactly,
+    # so we will use a couple of generic links for them if needed.
+    "Rice - Leaf Blast": "https://upload.wikimedia.org/wikipedia/commons/e/ea/Magnaporthe_grisea.jpg",
+}
+
+def verify_all_plants():
+    print("Downloading & Testing PlantVillage Images against HuggingFace Engine...")
+    print("API Endpoint:", API_URL)
+    print("-" * 60)
+    
+    results = []
+
+    for name, url in TEST_IMAGES.items():
+        try:
+            print(f"Downloading {name}...")
+            # We use urllib to handle spaces in URLs properly if any
+            req = urllib.request.Request(url, headers={'User-Agent': 'Mozilla/5.0'})
+            with urllib.request.urlopen(req) as response:
+                img_data = response.read()
+            
+            # Post to HuggingFace
+            print(f"Testing {name} on HuggingFace Space...")
+            files = {'file': (f'{name.replace(" ", "_")}.jpg', img_data, 'image/jpeg')}
+            
+            # Record time for performance metrics
+            start_time = time.time()
+            resp = requests.post(API_URL, files=files, timeout=40)
+            elapsed = time.time() - start_time
+            
+            if resp.status_code == 200:
+                result = resp.json()
+                status_icon = "PASS" if result.get('class') != "UNKNOWN" else "WARN"
+                print(f"{status_icon} RESULT for {name} ({elapsed:.1f}s):")
+                print(f"   ► Class Predicted: {result.get('class')}")
+                print(f"   ► Confidence: {result.get('confidence', 0)*100:.1f}%")
+                print(f"   ► AI Details: {result.get('ai_details')}")
+                
+                results.append({
+                    "Target": name,
+                    "Predicted": result.get('class'),
+                    "Confidence": f"{result.get('confidence', 0)*100:.1f}%",
+                    "Expert": result.get('ai_details')
+                })
+                
+            else:
+                print(f"FAILED for {name}: {resp.status_code} - {resp.text}")
+                
+        except Exception as e:
+            print(f"Error testing {name} (Link might be dead): {e}")
+        print("-" * 60)
+        time.sleep(1) # Prevent rate limiting
+
+    print("\n\n SUMMARY REPORT:")
+    for r in results:
+        match = "MATCH" if r['Predicted'].lower().split('-')[0].strip() in r['Target'].lower() else "CONFLICT"
+        print(f"Target: {r['Target']:<25} | Predicted: {r['Predicted']:<25} | {r['Confidence']} | {match}")
+
+if __name__ == "__main__":
+    verify_all_plants()

test_specialist_local.py

@@ -0,0 +1,55 @@
+from transformers import AutoModelForImageClassification
+import torch
+import numpy as np
+from PIL import Image
+import requests
+import io
+import urllib.request
+
+def test_local():
+    print("Loading MobileNetV2 locally...")
+    model_name = "linkanjarad/mobilenet_v2_1.0_224-plant-disease-identification"
+    model = AutoModelForImageClassification.from_pretrained(model_name).to(torch.float32)
+    model.eval()
+
+    print("Fetching sample Tomato Septoria image...")
+    url = "https://raw.githubusercontent.com/spMohanty/PlantVillage-Dataset/master/raw/color/Tomato___Septoria_leaf_spot/0a555f63-bf03-4958-8993-e1932b8dce9f___JR_Sept.L.S%208395.JPG"
+    
+    req = urllib.request.Request(url, headers={'User-Agent': 'Mozilla/5.0'})
+    try:
+        with urllib.request.urlopen(req) as response:
+            img_data = response.read()
+        image = Image.open(io.BytesIO(img_data)).convert("RGB").resize((224, 224))
+    except Exception as e:
+        print("Could not download image. Creating a synthetic green image with black spots...")
+        image = Image.new('RGB', (224, 224), color=(30, 150, 30))
+        # draw spots
+        for _ in range(100):
+            x, y = np.random.randint(0, 224, 2)
+            for i in range(3):
+                for j in range(3):
+                    if x+i<224 and y+j<224:
+                        image.putpixel((x+i, y+j), (50, 20, 0))
+
+    img_array = np.array(image).astype(np.float32) / 255.0
+    mean = np.array([0.5, 0.5, 0.5])
+    std = np.array([0.5, 0.5, 0.5])
+    img_array = (img_array - mean) / std
+    img_array = img_array.transpose(2, 0, 1)
+    inputs = torch.tensor(img_array, dtype=torch.float32).unsqueeze(0)
+
+    with torch.no_grad():
+        outputs = model(inputs)
+    
+    probs = torch.nn.functional.softmax(outputs.logits, dim=-1)[0]
+    
+    print("\n--- TOP 3 PREDICTIONS ---")
+    top3 = torch.topk(probs, 3)
+    for i in range(3):
+        idx = top3.indices[i].item()
+        score = top3.values[i].item()
+        label = model.config.id2label[idx]
+        print(f"{i+1}. {label} : {score*100:.2f}%")
+
+if __name__ == "__main__":
+    test_local()