add inference

inference/predictions.py

@@ -0,0 +1,50 @@
+from processor.tabular_processor import TabularProcessor
+from processor.image_processor import ImageProcessor
+from processor.text_processor import TextProcessor
+
+from utils.model_loader import EnhancedFusionModel
+from utils.dataset import InferenceDataset
+
+import torch
+from torch.utils.data import DataLoader
+
+from config import IMAGE_MODEL_NAME, TEXT_MODEL_NAME, TABULAR_DIM, IMAGE_DIM, TEXT_DIM, DEVICE, MODEL_PATH, SCALER_PATH
+
+class MultimodalPredictor:
+    def __init__(self, model_path=MODEL_PATH):
+        self.tabular_processor = TabularProcessor()
+        self.image_processor = ImageProcessor(model_name=IMAGE_MODEL_NAME)
+        self.text_processor = TextProcessor(model_name=TEXT_MODEL_NAME)
+
+        self.model = EnhancedFusionModel(
+            tab_dim=TABULAR_DIM,
+            img_dim=IMAGE_DIM,
+            text_dim=TEXT_DIM
+        )
+
+        self.model.load_state_dict(torch.load(model_path, map_location=DEVICE))
+        self.model.to(DEVICE)
+        self.model.eval()
+
+    def predict(self, tabular_data, image_path, text_data, batch_size=1):
+        tabular_processed = self.tabular_processor.process(tabular_data)
+        image_processed = self.image_processor.process(image_path)
+        text_processed = self.text_processor.process(text_data)
+
+        dataset = InferenceDataset(tabular_data=tabular_processed, image_features=image_processed, text_features=text_processed)
+        dataloader = DataLoader(dataset, batch_size=batch_size, shuffle=False)
+
+        predictions = []
+        with torch.no_grad():
+            for batch in dataloader:
+                tab_batch, img_batch, text_batch = [b.to(DEVICE) for b in batch]
+                outputs = self.model(tab_batch, img_batch, text_batch)
+                predictions.extend(outputs.cpu().numpy().flatten().tolist())
+
+        predictions = [pred * 1000000 for pred in predictions]
+        return predictions
+
+
+
+
+