Create script_fuse_with_baseline.py

script_fuse_with_baseline.py

@@ -0,0 +1,123 @@
+import pandas as pd
+import numpy as np
+import onnxruntime as ort
+import os
+from tqdm import tqdm
+import timm
+import torchvision.transforms as T
+from PIL import Image
+import torch
+import torch.nn as nn
+
+def is_gpu_available():
+    """Check if the python package `onnxruntime-gpu` is installed."""
+    return torch.cuda.is_available()
+
+
+class PytorchWorker:
+    """Run inference using ONNX runtime."""
+
+    def __init__(self, model_path: str, model_name: str, number_of_categories: int = 1605):
+
+        def _load_model(model_name, model_path):
+
+            print("Setting up Pytorch Model")
+            self.device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
+            print(f"Using devide: {self.device}")
+
+            model = timm.create_model(model_name, num_classes=number_of_categories, pretrained=False)
+
+            # if not torch.cuda.is_available():
+            #     model_ckpt = torch.load(model_path, map_location=torch.device("cpu"))
+            # else:
+            #     model_ckpt = torch.load(model_path)
+
+            model_ckpt = torch.load(model_path, map_location=self.device)
+            model.load_state_dict(model_ckpt, strict=False)
+            msg = model.load_state_dict(model_ckpt, strict=False)
+            print("load_state_dict: ", msg)
+            # num_features = model.get_classifier().in_features
+            # model.classifier = nn.Linear(num_features, number_of_categories)
+
+            return model.to(self.device).eval()
+
+        self.model = _load_model(model_name, model_path)
+
+        self.transforms = T.Compose([T.Resize((299, 299)),
+                                     T.ToTensor(),
+                                     T.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])])
+                                    #  T.Normalize([0.5, 0.5, 0.5], [0.5, 0.5, 0.5])])
+
+
+    def predict_image(self, image: np.ndarray) -> list:
+        """Run inference using ONNX runtime.
+        :param image: Input image as numpy array.
+        :return: A list with logits and confidences.
+        """
+        self.model.eval()
+        outputs = self.model(self.transforms(image).unsqueeze(0).to(self.device))
+        return outputs.cpu()  # Convert tensor to list
+
+
+def make_submission(test_metadata, model_path, model_path2, model_name, model_name2, output_csv_path="./submission.csv", images_root_path="/tmp/data/private_testset"):
+    """Make submission with given """
+    model = PytorchWorker(model_path, model_name, number_of_categories=1604)
+    model2 = PytorchWorker(model_path2, model_name2)
+    
+    predictions = []
+    correct_max_values = []
+    incorrect_max_values = []
+    
+    for _, row in tqdm(test_metadata.iterrows(), total=len(test_metadata)):
+        image_path = os.path.join(images_root_path, row.image_path)
+        test_image = Image.open(image_path).convert("RGB")
+        
+        outputs = model.predict_image(test_image)
+        outputs2 = model2.predict_image(test_image)
+        
+        # max_value = torch.max(outputs+outputs2)
+        _, preds = torch.max(outputs, 1) # baseline
+        _, preds2 = torch.max(outputs2, 1) # 1.4.3
+        
+        pred_class_id = preds.tolist()
+        pred_class_id2 = preds2.tolist()
+        # max_value2 = torch.max(outputs2)
+        
+        pred_class_id = pred_class_id[0] if pred_class_id2[0] != 1604 else -1
+        
+        predictions.append(pred_class_id)
+
+    test_metadata["class_id"] = predictions
+
+    user_pred_df = test_metadata.drop_duplicates("observation_id", keep="first")
+    user_pred_df[["observation_id", "class_id"]].to_csv(output_csv_path, index=None)
+
+
+if __name__ == "__main__":
+
+    import zipfile
+
+    with zipfile.ZipFile("/tmp/data/private_testset.zip", 'r') as zip_ref:
+        zip_ref.extractall("/tmp/data")
+
+    # MODEL_PATH = './efficientnet_b3_epoch_9_delete_pre.pth' # "./efficientnet_b3_epoch_9.pth"
+    # MODEL_PATH = './efficientnet_b3_epoch_24_trick1.2.3_0.6067.pth'
+    # MODEL_PATH = './efficientnet_b3_epoch_10_trick1.2.4_0.6016.pth'
+    # MODEL_PATH = './efficientnet_b3_epoch_3_trick1.2.3_a0.6067_l5.6311.pth'
+    # MODEL_PATH = './efficientnet_b3_epoch_21_trick1.2.5_a0.7237_l17.1662.pth'
+    # MODEL_PATH = './efficientnet_b3_epoch_21_trcik1.5.2.pth'
+    # MODEL_PATH = './efficientnet_b3_epoch_28_1.4.3.pth'
+    MODEL_PATH = './pytorch_model.bin'
+    MODEL_PATH2 = './efficientnet_b3_epoch_28_1.4.3.pth'
+    MODEL_NAME = "tf_efficientnet_b1_ap"
+    MODEL_NAME2 = "tf_efficientnet_b3_ns"
+    metadata_file_path = "./FungiCLEF2024_TestMetadata.csv"
+    test_metadata = pd.read_csv(metadata_file_path)
+
+    make_submission(
+        test_metadata=test_metadata,
+        model_path=MODEL_PATH,
+        model_path2=MODEL_PATH2,
+        model_name=MODEL_NAME,
+        model_name2=MODEL_NAME2
+    )