Update app.py

app.py

@@ -4,6 +4,10 @@ import torch.nn as nn
 import sys
 import os
 
+from fastapi import FastAPI, UploadFile, File
+from fastapi.middleware.cors import CORSMiddleware
+from fastapi.responses import JSONResponse
+
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
 
 from models.audio_model import FrozenExtractor, LinearProjection, AGRU, PersonalityRegressor
@@ -98,7 +102,6 @@ else:
     proj_state = {k[len("proj."):]: v for k, v in audio_state.items() if k.startswith("proj.")}
     agru_state = {k[len("agru."):]: v for k, v in audio_state.items() if k.startswith("agru.")}
     reg_state  = {k[len("regressor."):]: v for k, v in audio_state.items() if k.startswith("regressor.")}
-    
     if dt_state:
         deep_transformer.load_state_dict(dt_state)
         proj.load_state_dict(proj_state)
@@ -114,56 +117,76 @@ print("⏳ Loading face detector...")
 face_detector = _load_face_detector()
 print("✓ Face detector ready")
 
-# ── Inference ──────────────────────────────────────────────────────────────────
-def predict(video_path):
+
+# ── CORE LOGIC HELPER ──────────────────────────────────────────────────────────
+def process_video_bytes(video_bytes):
+    visual_preds, visual_emb = get_visual_embedding(video_bytes, visual_model, DEVICE, face_detector)
+    audio_preds, audio_emb = get_audio_embedding(video_bytes, frozen_ext, deep_transformer, proj, agru, regressor, DEVICE)
+    fusion_preds = get_fusion_prediction(audio_emb, visual_emb, fusion_model, DEVICE)
+    results = format_predictions(fusion_preds, visual_preds, audio_preds)
+    return results
+
+
+# ── FASTAPI SETUP (FOR REACT) ──────────────────────────────────────────────────
+app = FastAPI()
+
+# THE MAGIC BULLET FOR CORS: allow_origins=["*"] MUST HAVE allow_credentials=False
+app.add_middleware(
+    CORSMiddleware,
+    allow_origins=["*"],
+    allow_credentials=False, 
+    allow_methods=["*"],
+    allow_headers=["*"],
+)
+
+@app.post("/api/predict")
+async def api_predict(video: UploadFile = File(...)):
+    """This route is exclusively for your React app to call via Axios"""
+    try:
+        video_bytes = await video.read()
+        results = process_video_bytes(video_bytes)
+        return JSONResponse(content=results)
+    except Exception as e:
+        import traceback
+        return JSONResponse(content={"error": str(e), "trace": traceback.format_exc()}, status_code=500)
+
+
+# ── GRADIO UI SETUP (FOR MANUAL TESTING) ───────────────────────────────────────
+def gradio_predict(video_path):
     if video_path is None:
         return None, "⚠️ Please upload a video first."
-
     try:
         with open(video_path, "rb") as f:
             video_bytes = f.read()
-
-        visual_preds, visual_emb = get_visual_embedding(video_bytes, visual_model, DEVICE, face_detector)
-        audio_preds, audio_emb = get_audio_embedding(
-            video_bytes, frozen_ext, deep_transformer, proj, agru, regressor, DEVICE
-        )
-        fusion_preds = get_fusion_prediction(audio_emb, visual_emb, fusion_model, DEVICE)
-
-        results = format_predictions(fusion_preds, visual_preds, audio_preds)
-        fusion = results["fusion"]
         
+        results = process_video_bytes(video_bytes)
+        
+        fusion = results["fusion"]
         md = "## 🧠 Fusion Results (Main)\n\n"
         for trait, score in fusion.items():
             bar = "█" * int(score / 5) + "░" * (20 - int(score / 5))
             md += f"**{trait}** `{score:.1f}%`  {bar}\n\n"
-
         return results, md
-
     except Exception as e:
         import traceback
         return None, f"❌ Error: {str(e)}\n\n```{traceback.format_exc()}```"
 
-# ── Gradio UI ──────────────────────────────────────────────────────────────────
-with gr.Blocks(title="Personality Prediction API") as demo:
-
-    gr.Markdown("# 🧠 Personality Prediction\nUpload a video to predict **OCEAN** personality traits.")
-
+with gr.Blocks(title="Personality Prediction API", theme=gr.themes.Soft()) as demo:
+    gr.Markdown("# 🧠 Personality Prediction")
     with gr.Row():
         with gr.Column(scale=1):
             video_input = gr.Video(label="📹 Upload Video")
             predict_btn = gr.Button("🔍 Predict Personality", variant="primary", size="lg")
-
         with gr.Column(scale=1):
             summary_out = gr.Markdown(label="Results Summary")
-
     with gr.Accordion("📊 Full Results", open=False):
         json_out = gr.JSON(label="Raw Scores (%)")
 
     predict_btn.click(
-        fn=predict,
+        fn=gradio_predict,
         inputs=video_input,
-        outputs=[json_out, summary_out],
-        api_name="predict" # This creates the API endpoint
+        outputs=[json_out, summary_out]
     )
 
-demo.launch()
+# Mount Gradio over the FastAPI app
+app = gr.mount_gradio_app(app, demo, path="/")