fix: paste bad algo :(

Former-commit-id: 23df9b5127d0abd3cc9ac77f56b0a1755b28eb00 [formerly 30326f6dd6eace91f4e0e463eab3fdb2083fb38e]
Former-commit-id: 23cbd859a96b5cf0cc361b71b32b54f54edb321b

comp.ipynb.REMOVED.git-id

@@ -1 +1 @@
-eb93e8ed23714201f3e83001ee4821b791b01a53
+946598e599580eb39a7de79d3607416538bc0019

src/train.py

@@ -18,7 +18,7 @@ from utils.paste import RandomPaste
 
 CHECKPOINT_DIR = Path("./checkpoints/")
 DIR_TRAIN_IMG = Path("/home/lilian/data_disk/lfainsin/val2017")
-DIR_VALID_IMG = Path("/home/lilian/data_disk/lfainsin/val2017/")
+DIR_VALID_IMG = Path("/home/lilian/data_disk/lfainsin/smolval2017/")
 DIR_SPHERE_IMG = Path("/home/lilian/data_disk/lfainsin/spheres/Images/")
 DIR_SPHERE_MASK = Path("/home/lilian/data_disk/lfainsin/spheres/Masks/")
 
@@ -88,6 +88,9 @@ def main():
     device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
     logging.info(f"Using device {device}")
 
+    # enable cudnn benchmarking
+    torch.backends.cudnn.benchmark = True
+
     # 0. Create network
     net = UNet(n_channels=3, n_classes=args.classes)
     logging.info(
@@ -131,7 +134,7 @@ def main():
     ds_valid = SphereDataset(image_dir=DIR_VALID_IMG, transform=tf_valid)
 
     # 3. Create data loaders
-    loader_args = dict(batch_size=args.batch_size, num_workers=5, pin_memory=True)
+    loader_args = dict(batch_size=args.batch_size, num_workers=6, pin_memory=True)
     train_loader = DataLoader(ds_train, shuffle=True, **loader_args)
     val_loader = DataLoader(ds_valid, shuffle=False, drop_last=True, **loader_args)
 
@@ -141,6 +144,7 @@ def main():
     grad_scaler = torch.cuda.amp.GradScaler(enabled=args.amp)
     criterion = nn.BCEWithLogitsLoss()
 
+    # setup wandb
     wandb.init(
         project="U-Net-tmp",
         config=dict(
@@ -150,6 +154,7 @@ def main():
             amp=args.amp,
         ),
     )
+    wandb.save(f"{CHECKPOINT_DIR}/*")
 
     logging.info(
         f"""Starting training:

src/utils/paste.py

@@ -56,6 +56,13 @@ class RandomPaste(A.DualTransform):
 
         return np.asarray(mask.convert("L"))
 
+    @staticmethod
+    def overlap(positions, x1, y1, w, h):
+        for x2, y2 in positions:
+            if x1 + w >= x2 and x1 <= x2 + w and y1 + h >= y2 and y1 <= y2 + h:
+                return True
+        return False
+
     def get_params_dependent_on_targets(self, params):
         # choose a random image inside the image folder
         filename = rd.choice(os.listdir(self.path_paste_img_dir))
@@ -107,13 +114,13 @@ class RandomPaste(A.DualTransform):
 
         # generate some positions
         positions = []
-        while len(positions) <= rd.randint(1, self.nb):
+        NB = rd.randint(1, self.nb)
+        while len(positions) <= NB:
             x = rd.randint(0, target_shape[0] - paste_shape[0])
             y = rd.randint(0, target_shape[1] - paste_shape[1])
 
             # check for overlapping
-            for xo, yo in positions:
-                if (x <= xo + paste_shape[0]) and (y <= yo + paste_shape[1]):
+            if RandomPaste.overlap(positions, x, y, paste_shape[0], paste_shape[1]):
                 continue
 
             positions.append((x, y))