modifycapcha

main.py

@@ -1,81 +1,256 @@
 import os
+import sys
+import json
+import random
+import shutil
+import hashlib
 import uuid
+from typing import List
+import base64
+from io import BytesIO
 import time
-import shutil
-import json
-import requests
 import threading
+import numpy as np
+import torch
+import torch.nn as nn
+from PIL import Image, ImageOps
+from matplotlib import cm
+import requests
 
-from fastapi import FastAPI, Request, Form, UploadFile, File
-from fastapi.responses import RedirectResponse
-from fastapi.staticfiles import StaticFiles
+import cv2
+from fastapi import FastAPI, File, UploadFile, Form, Request
+from fastapi.responses import HTMLResponse, RedirectResponse
 from fastapi.templating import Jinja2Templates
+from fastapi.staticfiles import StaticFiles
 
-from dotenv import load_dotenv
-load_dotenv()
+# ===============================
+# DOWNLOAD MODEL FROM HF IF NEEDED
+# ===============================
+HF_MODEL_URL = "https://huggingface.co/qqqqqqat/densenet_wangchan/resolve/main/best_fusion_densenet.pth"
+LOCAL_MODEL_PATH = "models/densenet/best_fusion_densenet.pth"
 
-from your_model_module import process_with_ai_model  # ← เปลี่ยนเป็นของคุณ
+def download_model_if_needed():
+    if not os.path.exists(LOCAL_MODEL_PATH):
+        print("📥 Downloading model from HuggingFace...")
+        os.makedirs(os.path.dirname(LOCAL_MODEL_PATH), exist_ok=True)
+        response = requests.get(HF_MODEL_URL)
+        with open(LOCAL_MODEL_PATH, "wb") as f:
+            f.write(response.content)
+        print("✅ Model downloaded!")
 
-app = FastAPI()
+# =================================
+
+sys.path.append(os.path.abspath(os.path.dirname(__file__)))
+from models.densenet.preprocess.preprocessingwangchan import get_tokenizer, get_transforms
+from models.densenet.train_densenet_only import DenseNet121Classifier
+from models.densenet.train_text_only import TextClassifier
+
+torch.manual_seed(42); np.random.seed(42); random.seed(42)
+
+FUSION_LABELMAP_PATH = "models/densenet/label_map_fusion_densenet.json"
+
+with open(FUSION_LABELMAP_PATH, "r", encoding="utf-8") as f:
+    label_map = json.load(f)
+
+class_names = [label for label, _ in sorted(label_map.items(), key=lambda x: x[1])]
+NUM_CLASSES = len(class_names)
+
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+print("🧠 Using:", device)
+
+# ==========================
+# MODEL DEFINITION
+# ==========================
+
+class FusionDenseNetText(nn.Module):
+    def __init__(self, num_classes, dropout=0.3):
+        super().__init__()
+        self.image_model = DenseNet121Classifier(num_classes=num_classes)
+        self.text_model  = TextClassifier(num_classes=num_classes)
+        self.fusion = nn.Sequential(
+            nn.Linear(num_classes * 2, 128), nn.ReLU(),
+            nn.Dropout(dropout), nn.Linear(128, num_classes)
+        )
+
+    def forward(self, image, input_ids, attention_mask):
+        logits_img = self.image_model(image)
+        logits_txt = self.text_model(input_ids, attention_mask)
+        fused_in   = torch.cat([logits_img, logits_txt], dim=1)
+        fused_out  = self.fusion(fused_in)
+        return fused_out, logits_img, logits_txt
+
+# ====================================
+# LOAD MODEL
+# ====================================
+print("🔄 Loading model...")
+download_model_if_needed()
+
+fusion_model = FusionDenseNetText(num_classes=NUM_CLASSES).to(device)
+fusion_model.load_state_dict(torch.load(LOCAL_MODEL_PATH, map_location=device))
+fusion_model.eval()
+print("✅ Model loaded!")
+
+tokenizer = get_tokenizer()
+transform = get_transforms((224, 224))
+
+# ====================================
+# Helper for GradCAM
+# ====================================
+
+def _find_last_conv2d(mod: torch.nn.Module):
+    last = None
+    for m in mod.modules():
+        if isinstance(m, torch.nn.Conv2d): last = m
+    return last
+
+def compute_gradcam_overlay(img_pil, image_tensor, target_idx):
+    img_branch = fusion_model.image_model
+    target_layer = _find_last_conv2d(img_branch)
+    if target_layer is None:
+        return None
+
+    activations, gradients = [], []
+
+    def fwd(_, __, o): activations.append(o)
+    def bwd(_, gin, gout): gradients.append(gout[0])
+
+    h1 = target_layer.register_forward_hook(fwd)
+    h2 = target_layer.register_full_backward_hook(bwd)
+
+    try:
+        img_branch.zero_grad()
+        logits_img = img_branch(image_tensor)
+        score = logits_img[0, target_idx]
+        score.backward()
+
+        act = activations[-1].detach()[0]
+        grad = gradients[-1].detach()[0]
+        weights = torch.mean(grad, dim=(1, 2))
+
+        cam = torch.relu(torch.sum(weights[:, None, None] * act, dim=0))
+        cam -= cam.min()
+        cam /= (cam.max() + 1e-8)
 
+        cam_img = Image.fromarray((cam.cpu().numpy() * 255).astype(np.uint8)).resize(img_pil.size)
+        heatmap = cm.get_cmap("jet")(cam_img)[:, :, :3]
+
+        img_np = np.asarray(img_pil.convert("RGB")).astype(np.float32) / 255.0
+        overlay = (0.6 * img_np + 0.4 * heatmap)
+        return np.clip(overlay * 255, 0, 255).astype(np.uint8)
+
+    finally:
+        h1.remove()
+        h2.remove()
+        img_branch.zero_grad()
+
+# ====================================
+# FASTAPI SETUP
+# ====================================
+
+app = FastAPI()
 templates = Jinja2Templates(directory="templates")
 app.mount("/static", StaticFiles(directory="static"), name="static")
+os.makedirs("uploads", exist_ok=True)
 
+EXPIRATION_MINUTES = 10
 results_cache = {}
 cache_lock = threading.Lock()
 
-
 SYMPTOM_MAP = {
-    "pain_touch": "เจ็บเมื่อโดนแผล",
-    "eat_spicy": "กินเผ็ดแสบ",
-    "drink_alcohol": "ดื่มเหล้า",
+    "noSymptoms": "ไม่มีอาการ",
+    "drinkAlcohol": "ดื่มเหล้า",
     "smoking": "สูบบุหรี่",
-    "betel_chew": "เคี้ยวหมาก",
-    "wipe_off": "เช็ดออกได้",
-    "no_symptoms": "ไม่มีอาการ",
+    "chewBetelNut": "เคี้ยวหมาก",
+    "eatSpicyFood": "กินเผ็ดแสบ",
+    "wipeOff": "เช็ดออกได้",
+    "alwaysHurts": "เจ็บเมื่อโดนแผล"
 }
 
+# ====================================
+# AI MAIN PROCESS
+# ====================================
+
+def process_with_ai_model(image_path, prompt_text):
+    try:
+        image_pil = Image.open(image_path)
+        image_pil = ImageOps.exif_transpose(image_pil)
+        image_pil = image_pil.convert("RGB")
+
+        tensor = transform(image_pil).unsqueeze(0).to(device)
+        enc = tokenizer(prompt_text, return_tensors="pt", padding="max_length",
+                        truncation=True, max_length=128)
+
+        ids = enc["input_ids"].to(device)
+        mask = enc["attention_mask"].to(device)
+
+        with torch.no_grad():
+            fused_logits, _, _ = fusion_model(tensor, ids, mask)
+            probs = torch.softmax(fused_logits, dim=1)[0].cpu().numpy()
+
+        pred_idx = int(np.argmax(probs))
+        pred_label = class_names[pred_idx]
+        confidence = float(probs[pred_idx]) * 100
+
+        gradcam_overlay_np = compute_gradcam_overlay(image_pil, tensor, pred_idx)
+
+        def img64(img):
+            buf = BytesIO()
+            img.save(buf, format="JPEG")
+            return base64.b64encode(buf.getvalue()).decode()
+
+        original_b64 = img64(image_pil)
+
+        if gradcam_overlay_np is not None:
+            grad_pil = Image.fromarray(gradcam_overlay_np)
+            gradcam_b64 = img64(grad_pil)
+        else:
+            gradcam_b64 = original_b64
+
+        return original_b64, gradcam_b64, pred_label, f"{confidence:.2f}"
 
-@app.get("/detect")
+    except Exception as e:
+        print("❌ AI error:", e)
+        return None, None, "Error", "0.00"
+
+# ====================================
+# ROUTES
+# ====================================
+
+@app.get("/", response_class=RedirectResponse)
+def root():
+    return RedirectResponse("/detect")
+
+@app.get("/detect", response_class=HTMLResponse)
 def detect_page(request: Request):
     return templates.TemplateResponse("detect.html", {"request": request})
 
-
-@app.get("/results/{result_id}")
+@app.get("/results/{result_id}", response_class=HTMLResponse)
 def show_results(request: Request, result_id: str):
     with cache_lock:
-        result = results_cache.get(result_id)
+        item = results_cache.get(result_id)
 
-    if not result:
-        return templates.TemplateResponse("detect.html", {"request": request, "error": "ไม่พบผลลัพธ์"})
+    if not item:
+        return RedirectResponse("/detect")
 
-    data = result["data"]
     return templates.TemplateResponse(
         "detect.html",
-        {
-            "request": request,
-            "image_b64_data": data["image_b64_data"],
-            "gradcam_b64_data": data["gradcam_b64_data"],
-            "name_out": data["name_out"],
-            "eva_output": data["eva_output"],
-        }
+        {"request": request, **item["data"]}
     )
 
+# ====================================
+# 🔥 FINAL /uploaded (รวมแล้ว + เพิ่ม CAPTCHA)
+# ====================================
 
-# ==============================================
-#       SINGLE UPLOADED ENDPOINT (FIXED)
-# ==============================================
 @app.post("/uploaded")
 async def handle_upload(
     request: Request,
     file: UploadFile = File(...),
-    checkboxes: list = Form([]),
+    checkboxes: List[str] = Form([]),
     symptom_text: str = Form(""),
-    cf_token: str = Form(alias="cf-turnstile-response")
+    cf_token: str = Form(default=None, alias="cf-turnstile-response")
 ):
-
-    # ===== CAPTCHA CHECK =====
     TURNSTILE_SECRET = os.getenv("TURNSTILE_SECRET")
+
     if not cf_token:
         return templates.TemplateResponse(
             "detect.html",
@@ -83,45 +258,63 @@ async def handle_upload(
             status_code=400
         )
 
-    verify_res = requests.post(
+    verify = requests.post(
         "https://challenges.cloudflare.com/turnstile/v0/siteverify",
         data={"secret": TURNSTILE_SECRET, "response": cf_token}
     ).json()
 
-    if not verify_res.get("success", False):
+    if not verify.get("success", False):
         return templates.TemplateResponse(
             "detect.html",
-            {"request": request, "error": "CAPTCHA ไม่ผ่านการตรวจสอบ กรุณาลองใหม่"},
+            {"request": request, "error": "CAPTCHA ไม่ผ่านการตรวจสอบ"},
             status_code=400
         )
 
-    # ===== SAVE IMAGE =====
     temp_path = os.path.join("uploads", f"{uuid.uuid4()}_{file.filename}")
+
     with open(temp_path, "wb") as buffer:
         shutil.copyfileobj(file.file, buffer)
 
-    selected_symptoms = {SYMPTOM_MAP.get(cb) for cb in checkboxes if SYMPTOM_MAP.get(cb)}
-    final_prompt = " ".join(selected_symptoms) if selected_symptoms else "ไม่มีอาการ"
+    selected = {SYMPTOM_MAP.get(cb) for cb in checkboxes if SYMPTOM_MAP.get(cb)}
 
-    # ===== AI MODEL PROCESS =====
-    image_b64, gradcam_b64, name_out, eva_output = process_with_ai_model(
-        image_path=temp_path,
-        prompt_text=final_prompt
-    )
+    parts = []
+
+    if "ไม่มีอาการ" in selected:
+        symptoms   = {"เจ็บเมื่อโดนแผล", "กินเผ็ดแสบ"}
+        lifestyles = {"ดื่มเหล้า", "สูบบุหรี่", "เคี้ยวหมาก"}
+        patterns   = {"เช็ดออกได้"}
+        specials   = {"ไม่มีอาการ"}
+
+        final_sel = (selected - symptoms) | (selected & (lifestyles | patterns | specials))
+        parts.append(" ".join(sorted(final_sel)))
+    elif selected:
+        parts.append(" ".join(sorted(selected)))
+
+    if symptom_text.strip():
+        parts.append(symptom_text.strip())
 
+    final_prompt = "; ".join(parts) if parts else "ไม่มีอาการ"
+
+    img_b64, grad_b64, lbl, conf = process_with_ai_model(temp_path, final_prompt)
     os.remove(temp_path)
 
-    # ===== STORE RESULT =====
     result_id = str(uuid.uuid4())
     with cache_lock:
         results_cache[result_id] = {
             "data": {
-                "image_b64_data": image_b64,
-                "gradcam_b64_data": gradcam_b64,
-                "name_out": name_out,
-                "eva_output": eva_output,
+                "image_b64_data": img_b64,
+                "gradcam_b64_data": grad_b64,
+                "name_out": lbl,
+                "eva_output": conf,
             },
-            "created_at": time.time()
+            "created_at": time.time(),
         }
 
-    return RedirectResponse(url=f"/results/{result_id}", status_code=303)
+    return RedirectResponse(f"/results/{result_id}", status_code=303)
+
+# ====================================
+
+if __name__ == "__main__":
+    port = int(os.environ.get("PORT", 8000))
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=port)

templates/detect.html

@@ -1,10 +1,11 @@
+
 <!DOCTYPE html>
 <html lang="th">
 <head>
     <meta charset="UTF-8">
     <title>Detect Oral Lesion</title>
-    <link rel="stylesheet" href="/static/bootstrap.min.css">
-    <link rel="stylesheet" href="/static/style.css">
+    <link href="/static/bootstrap.min.css" rel="stylesheet">
+    <link href="/static/style.css" rel="stylesheet">
 </head>
 
 <body class="bg-light">
@@ -24,19 +25,20 @@
         </div>
 
         <div class="mb-3">
-            <label class="form-label fw-bold">เลือกอาการ/ปัจจัยเสี่ยง</label>
+            <label class="form-label fw-bold">เลือกอาการหรือปัจจัยร่วม</label>
+
             <div class="form-check">
-                <input class="form-check-input" type="checkbox" name="checkboxes" value="pain_touch">
+                <input class="form-check-input" type="checkbox" name="checkboxes" value="alwaysHurts">
                 <label class="form-check-label">เจ็บเมื่อโดนแผล</label>
             </div>
 
             <div class="form-check">
-                <input class="form-check-input" type="checkbox" name="checkboxes" value="eat_spicy">
+                <input class="form-check-input" type="checkbox" name="checkboxes" value="eatSpicyFood">
                 <label class="form-check-label">กินเผ็ดแสบ</label>
             </div>
 
             <div class="form-check">
-                <input class="form-check-input" type="checkbox" name="checkboxes" value="drink_alcohol">
+                <input class="form-check-input" type="checkbox" name="checkboxes" value="drinkAlcohol">
                 <label class="form-check-label">ดื่มเหล้า</label>
             </div>
 
@@ -46,19 +48,20 @@
             </div>
 
             <div class="form-check">
-                <input class="form-check-input" type="checkbox" name="checkboxes" value="betel_chew">
+                <input class="form-check-input" type="checkbox" name="checkboxes" value="chewBetelNut">
                 <label class="form-check-label">เคี้ยวหมาก</label>
             </div>
 
             <div class="form-check">
-                <input class="form-check-input" type="checkbox" name="checkboxes" value="wipe_off">
+                <input class="form-check-input" type="checkbox" name="checkboxes" value="wipeOff">
                 <label class="form-check-label">เช็ดออกได้</label>
             </div>
 
             <div class="form-check">
-                <input class="form-check-input" type="checkbox" name="checkboxes" value="no_symptoms">
+                <input class="form-check-input" type="checkbox" name="checkboxes" value="noSymptoms">
                 <label class="form-check-label">ไม่มีอาการ</label>
             </div>
+
         </div>
 
         <div class="mb-3">
@@ -66,9 +69,9 @@
             <textarea class="form-control" name="symptom_text" rows="3"></textarea>
         </div>
 
-        <!-- ===========================
-        CLOUD FLARE TURNSTILE CAPTCHA
-        ============================ -->
+        <!-- ============================= -->
+        <!-- Cloudflare Turnstile CAPTCHA -->
+        <!-- ============================= -->
         <div class="cf-turnstile mt-3 mb-3"
              data-sitekey="0x4AAAAAACEfyPjr3pfV21Mm"
              data-callback="onTurnstileSuccess">
@@ -85,12 +88,11 @@
         </script>
 
         <div class="text-center mt-4">
-            <button class="btn btn-primary px-4 py-2" type="submit">Submit</button>
+            <button type="submit" class="btn btn-primary px-4 py-2">Submit</button>
         </div>
 
     </form>
 
-    <!-- Modal แสดงผล -->
     {% if image_b64_data %}
     <div class="modal fade show d-block" tabindex="-1">
         <div class="modal-dialog modal-lg">
@@ -98,17 +100,19 @@
                 <h4 class="text-center">ผลการตรวจ</h4>
                 <p class="text-center fw-bold">{{ name_out }}</p>
 
-                <img src="data:image/png;base64,{{ image_b64_data }}" class="img-fluid mb-3">
-                <img src="data:image/png;base64,{{ gradcam_b64_data }}" class="img-fluid">
+                <img src="data:image/jpeg;base64,{{ image_b64_data }}" class="img-fluid mb-3" />
+                <img src="data:image/jpeg;base64,{{ gradcam_b64_data }}" class="img-fluid" />
 
-                <p class="mt-3">{{ eva_output }}</p>
+                <p class="mt-3">{{ eva_output }}%</p>
 
                 <a href="/detect" class="btn btn-secondary mt-3">ตรวจใหม่</a>
             </div>
         </div>
     </div>
     {% endif %}
+
 </div>
 
 </body>
 </html>
+