mlflow + conda

.gitignore

@@ -175,4 +175,5 @@ ipython_config.py
 #   git rm -r .ipynb_checkpoints/
 
 football_dataset/
-venvium/
+venvium/
+lightning_logs/

MLProject

@@ -0,0 +1,11 @@
+name: iumKC
+
+conda_env: environment.yaml
+
+entry_points:
+  main:
+    command: "python train.py --train=true --save_model=true"
+  test:
+    parameters:
+      model_filepath: "model"
+    command: "python test.py --test=true --load_model=$model_filepath"

id2label.json

@@ -0,0 +1 @@
+{"0": "Goal Bar", "1": "Referee", "2": "Advertisement", "3": "Ground", "4": "Ball", "5": "Coaches & Officials", "6": "Team A", "7": "Team B", "8": "Goalkeeper A", "9": "Goalkeeper B", "10": "Audience"}

main.py

@@ -1,6 +1,9 @@
+import argparse
 import random
 
 from lightning import Trainer
+import lightning as L
+from lightning.pytorch.loggers import MLFlowLogger
 from sklearn.model_selection import train_test_split
 from torch.utils.data import DataLoader
 
@@ -12,7 +15,7 @@ from data import (
 from model import UNetLightning
 
 
-def main():
+def main(train=True, test=True, save_model=False, load_model=None):
     all_images, all_paths = get_data()
     image_train_paths, image_test_paths, label_train_paths, label_test_paths = (
         train_test_split(all_images, all_paths, test_size=0.2, random_state=42)
@@ -28,11 +31,13 @@ def main():
     test_dataset = FootballSegDataset(
         image_test_paths, label_test_paths, test_mean, test_std, test_label_mean, test_label_std)
 
-    train_loader = DataLoader(train_dataset, batch_size=32, shuffle=True)
-    test_loader = DataLoader(test_dataset, batch_size=32, shuffle=False)
+    train_loader = DataLoader(
+        train_dataset, batch_size=32, shuffle=True, num_workers=7, persistent_workers=True)
+    test_loader = DataLoader(test_dataset, batch_size=32,
+                             shuffle=False, num_workers=7, persistent_workers=True)
 
-    train_indices = random.sample(range(len(train_loader.dataset)), 5)
-    test_indices = random.sample(range(len(test_loader.dataset)), 5)
+    # train_indices = random.sample(range(len(train_loader.dataset)), 5)
+    # test_indices = random.sample(range(len(test_loader.dataset)), 5)
 
     # plot_random_images(train_indices, train_loader, "train")
     # plot_random_images(test_indices, test_loader, "test")
@@ -40,14 +45,32 @@ def main():
     # statistics.count_colors()
     # statistics.print_statistics()
 
+    mlFlowLogger = MLFlowLogger(
+        experiment_name="football-segmentation", tracking_uri="http://127.0.0.1:8080")
     model = UNetLightning(3, 3, learning_rate=1e-3)
-    trainer = Trainer(max_epochs=5, logger=True, log_every_n_steps=1)
-    trainer.fit(model, train_loader, test_loader)
+    trainer = Trainer(max_epochs=2, logger=mlFlowLogger, log_every_n_steps=1)
+    if train:
+        trainer.fit(model, train_loader, test_loader)
 
-    model.eval()
-    model.freeze()
-    trainer.test(model, test_loader)
+    if save_model:
+        model.save_hyperparameters()
+
+    if load_model:
+        model = UNetLightning.load_from_checkpoint(load_model)
+        model.eval()
+        model.freeze()
+
+    if test:
+        trainer.test(model, test_loader)
 
 
 if __name__ == "__main__":
-    main()
+    args = argparse.ArgumentParser()
+    args.add_argument("--seed", type=int, default=42)
+    args.add_argument("--train", type=bool, default=False)
+    args.add_argument("--test", type=bool, default=False)
+    args.add_argument("--save_model", type=bool, default=False)
+    args.add_argument("--load_model", type=str, default=None)
+    args = args.parse_args()
+    L.seed_everything(args.seed)
+    main(args.train, args.test, args.save_model, args.load_model)

model.py

@@ -47,14 +47,17 @@ class UNetLightning(L.LightningModule):
 
     def training_step(self, batch, batch_idx):
         x, y = batch
-        # print(max(x.flatten()), min(x.flatten()),
-        #       max(y.flatten()), min(y.flatten()))
         y_hat = self(x)
         loss = nn.CrossEntropyLoss()(y_hat, y)
         self.log("train_loss", loss)
         return loss
 
+    def log_hyperparameters(self):
+        for key, value in self.hparams.items():
+            self.logger.experiment.log_param(key, value)
+
     def configure_optimizers(self):
+        self.log_hyperparameters()
         optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)
         return optimizer
 
@@ -74,39 +77,3 @@ class UNetLightning(L.LightningModule):
         y_hat = self(x)
         loss = nn.CrossEntropyLoss()(y_hat, y)
         self.log("test_loss", loss)
-        # visualize
-        # if batch_idx == 0:
-        #     import matplotlib.pyplot as plt
-        #     import numpy as np
-        #     import torchvision.transforms as transforms
-        #     x = transforms.ToPILImage()(x[0])
-        #     y = transforms.ToPILImage()(y[0])
-        #     y_hat = transforms.ToPILImage()(y_hat[0])
-        #     plt.figure(figsize=(15, 15))
-        #     plt.subplot(131)
-        #     plt.imshow(np.array(x))
-        #     plt.title("Input")
-        #     plt.axis("off")
-        #     plt.subplot(132)
-        #     plt.imshow(np.array(y))
-        #     plt.title("Ground Truth")
-        #     plt.axis("off")
-        #     plt.subplot(133)
-        #     plt.imshow(np.array(y_hat))
-        #     plt.title("Prediction")
-        #     plt.axis("off")
-        #     plt.show()
-
-    # def on_test_epoch_end(self):
-    #     all_preds, all_labels = [], []
-    #     for output in self.trainer.predictions:
-    #         # predicted values
-    #         probs = list(output['logits'].cpu().detach().numpy())
-    #         labels = list(output['labels'].flatten().cpu().detach().numpy())
-    #         all_preds.extend(probs)
-    #         all_labels.extend(labels)
-
-    #     # save predictions and labels
-    #     import numpy as np
-    #     np.save('predictions.npy', all_preds)
-    #     np.save('labels.npy', all_labels)

requirements.txt

@@ -2,5 +2,5 @@ kaggle==1.6.6
 matplotlib==3.6.3
 Pillow==10.2.0
 scikit_learn==1.2.2
-torch==2.0.0+cu117
-torchvision==0.15.1+cu117
+# torch==2.0.0+cu117
+# torchvision==0.15.1+cu117