visual features handling

app.py

@@ -11,7 +11,7 @@ urls = [
 os.makedirs("data", exist_ok=True)
 
 # =========================
-# 2. LOAD LABELS FIRST
+# 2. LOAD LABELS
 # =========================
 import pandas as pd
 
@@ -32,21 +32,23 @@ train_labels = load_split("train_split_Depression_AVEC2017.csv")
 dev_labels = load_split("dev_split_Depression_AVEC2017.csv")
 
 ALL_REQUIRED_IDS = set(list(train_labels.keys()) + list(dev_labels.keys()))
-
 print("Total required participants:", len(ALL_REQUIRED_IDS))
 
 # =========================
-# 3. SELECTIVE EXTRACTION
+# 3. SELECTIVE EXTRACTION (FIXED)
 # =========================
 def extract_needed(zip_path):
     with zipfile.ZipFile(zip_path, "r") as zip_ref:
         for file in zip_ref.namelist():
 
-            # extract only participant folders we need
-            for pid in ALL_REQUIRED_IDS:
-                if f"{pid}_" in file:
-                    zip_ref.extract(file, "data")
-                    break
+            parts = file.split("/")
+            if len(parts) < 2:
+                continue
+
+            folder = parts[1] if parts[0].startswith("DAIC") else parts[0]
+
+            if any(folder.startswith(pid + "_") for pid in ALL_REQUIRED_IDS):
+                zip_ref.extract(file, "data")
 
 # =========================
 # 4. DOWNLOAD + EXTRACT
@@ -60,10 +62,10 @@ for i, url in enumerate(urls):
         for chunk in r.iter_content(8192):
             f.write(chunk)
 
-    print(f"Extracting required files from dataset {i+1}...")
+    print(f"Extracting dataset {i+1}...")
     extract_needed(zip_path)
 
-    os.remove(zip_path)  # 🔥 VERY IMPORTANT
+    os.remove(zip_path)
 
 # =========================
 # 5. GET PARTICIPANTS
@@ -77,10 +79,10 @@ def get_all_paths():
     return paths
 
 ALL_PATHS = get_all_paths()
-print("Extracted participant folders:", len(ALL_PATHS))
+print("Extracted participants:", len(ALL_PATHS))
 
 # =========================
-# 6. IMPORT LIBRARIES
+# 6. LIBRARIES
 # =========================
 import numpy as np
 import librosa
@@ -92,11 +94,11 @@ from transformers import AutoTokenizer, AutoModel
 from sklearn.metrics import accuracy_score, f1_score
 from sklearn.preprocessing import StandardScaler
 
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
 # =========================
 # 7. TEXT MODEL
 # =========================
-device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
-
 tokenizer = AutoTokenizer.from_pretrained("distilbert-base-uncased")
 bert = AutoModel.from_pretrained("distilbert-base-uncased").to(device)
 bert.eval()
@@ -131,19 +133,35 @@ def get_audio_features(folder):
     except:
         return np.zeros(40)
 
+# 🔥 FIXED VISUAL FUNCTION
 def get_visual_features(folder):
     feats = []
 
     for key in ["AUs", "pose", "gaze"]:
         try:
             file = [f for f in os.listdir(folder) if key in f][0]
-            df = pd.read_csv(os.path.join(folder, file))
+            path = os.path.join(folder, file)
 
-            cols = [c for c in df.columns if key.lower() in c.lower()]
-            df = df[cols]
+            df = pd.read_csv(path)
+
+            # keep only numeric columns
+            df = df.select_dtypes(include=[np.number])
+
+            # replace invalid values
+            df.replace(-100, np.nan, inplace=True)
+            df.fillna(0, inplace=True)
+
+            if df.shape[1] == 0:
+                feats.append(np.zeros(20))
+                continue
+
+            feat = np.concatenate([
+                df.mean().values,
+                df.std().values
+            ])
 
-            feat = np.concatenate([df.mean().values, df.std().values])
             feats.append(feat)
+
         except:
             feats.append(np.zeros(20))
 
@@ -173,6 +191,7 @@ Xt_test, Xa_test, Xv_test, y_test = build(dev_labels)
 
 print("Train size:", len(y_train))
 print("Test size:", len(y_test))
+print("Visual shape:", Xv.shape)
 
 # =========================
 # 10. NORMALIZE
@@ -196,6 +215,7 @@ class Model(nn.Module):
         self.t = nn.Sequential(nn.Linear(768,128), nn.ReLU())
         self.a = nn.Sequential(nn.Linear(40,32), nn.ReLU())
         self.v = nn.Sequential(nn.Linear(vdim,64), nn.ReLU())
+
         self.f = nn.Sequential(
             nn.Linear(224,64),
             nn.ReLU(),