feat: saving models with wandb

Former-commit-id: 994d061d49c8be9f680878d093b1e2b5710fa37c [formerly 3cadb0699c150421244c78511ee61b14262728d0]
Former-commit-id: 64b01c6ed73ba64a25d85acd65adc9b6c9cf126e

src/evaluate.py

@@ -20,18 +20,18 @@ def evaluate(net, dataloader, device):
             # forward, predict the mask
             with torch.inference_mode():
                 masks_pred = net(images)
-                masks_pred = (torch.sigmoid(masks_pred) > 0.5).float()
+                masks_pred_bin = (torch.sigmoid(masks_pred) > 0.5).float()
 
                 # compute the Dice score
-                dice_score += dice_coeff(masks_pred, masks_true, reduce_batch_first=False)
+                dice_score += dice_coeff(masks_pred_bin, masks_true, reduce_batch_first=False)
 
             # update progress bar
             pbar.update(images.shape[0])
 
     # save some images to wandb
-    table = wandb.Table(columns=["image", "mask", "prediction"])
-    for img, mask, pred in zip(images.to("cpu"), masks_true.to("cpu"), masks_pred.to("cpu")):
-        table.add_data(wandb.Image(img), wandb.Image(mask), wandb.Image(pred))
+    table = wandb.Table(columns=["id", "image", "mask", "prediction"])
+    for i, (img, mask, pred) in enumerate(zip(images.to("cpu"), masks_true.to("cpu"), masks_pred.to("cpu"))):
+        table.add_data(i, wandb.Image(img), wandb.Image(mask), wandb.Image(pred))
     wandb.log({"predictions_table": table}, commit=False)
 
     net.train()

src/predict.py

@@ -13,32 +13,6 @@ from unet import UNet
 from utils.utils import plot_img_and_mask
 
 
-def predict_img(net, full_img, device, scale_factor=1, out_threshold=0.5):
-    net.eval()
-    img = torch.from_numpy(BasicDataset.preprocess(full_img, scale_factor, is_mask=False))
-    img = img.unsqueeze(0)
-    img = img.to(device=device, dtype=torch.float32)
-
-    with torch.no_grad():
-        output = net(img)
-
-        if net.n_classes > 1:
-            probs = F.softmax(output, dim=1)[0]
-        else:
-            probs = torch.sigmoid(output)[0]
-
-        tf = transforms.Compose(
-            [transforms.ToPILImage(), transforms.Resize((full_img.size[1], full_img.size[0])), transforms.ToTensor()]
-        )
-
-        full_mask = tf(probs.cpu()).squeeze()
-
-    if net.n_classes == 1:
-        return (full_mask > out_threshold).numpy()
-    else:
-        return F.one_hot(full_mask.argmax(dim=0), net.n_classes).permute(2, 0, 1).numpy()
-
-
 def get_args():
     parser = argparse.ArgumentParser(
         description="Predict masks from input images",
@@ -95,6 +69,29 @@ def get_args():
     return parser.parse_args()
 
 
+def predict_img(net, full_img, device, scale_factor=1, out_threshold=0.5):
+    net.eval()
+    img = torch.from_numpy(BasicDataset.preprocess(full_img, scale_factor, is_mask=False))
+    img = img.unsqueeze(0)
+    img = img.to(device=device, dtype=torch.float32)
+
+    with torch.inference_mode():
+        output = net(img)
+
+        probs = torch.sigmoid(output)[0]
+
+        tf = transforms.Compose(
+            [transforms.ToPILImage(), transforms.Resize((full_img.size[1], full_img.size[0])), transforms.ToTensor()]
+        )
+
+        full_mask = tf(probs.cpu()).squeeze()
+
+    if net.n_classes == 1:
+        return (full_mask > out_threshold).numpy()
+    else:
+        return F.one_hot(full_mask.argmax(dim=0), net.n_classes).permute(2, 0, 1).numpy()
+
+
 def get_output_filenames(args):
     def _generate_name(fn):
         split = os.path.splitext(fn)
@@ -127,6 +124,7 @@ if __name__ == "__main__":
     logging.info("Model loaded!")
 
     for i, filename in enumerate(in_files):
+
         logging.info(f"\nPredicting image {filename} ...")
         img = Image.open(filename)
 

src/train.py

@@ -111,10 +111,11 @@ def main():
     # 1. Create transforms
     tf_train = A.Compose(
         [
-            A.Resize(500, 500),
+            A.Resize(512, 512),
             A.Flip(),
             A.ColorJitter(),
             RandomPaste(5, DIR_SPHERE_IMG, DIR_SPHERE_MASK),
+            A.GaussianBlur(),
             A.ISONoise(),
             A.ToFloat(max_value=255),
             ToTensorV2(),
@@ -122,7 +123,7 @@ def main():
     )
     tf_valid = A.Compose(
         [
-            A.Resize(500, 500),
+            A.Resize(512, 512),
             RandomPaste(5, DIR_SPHERE_IMG, DIR_SPHERE_MASK),
             A.ToFloat(max_value=255),
             ToTensorV2(),
@@ -154,7 +155,7 @@ def main():
             amp=args.amp,
         ),
     )
-    wandb.save(f"{CHECKPOINT_DIR}/*")
+    wandb.watch(net, log_freq=100)
 
     logging.info(
         f"""Starting training: