Fix top_left naming.

configs/benchmark.yaml

@@ -15,7 +15,7 @@ huggingface:
   image_res: [900, 1200]  # used during measurement
   rotate: True   # if measurement is upside-down
   alignment:
-    topright: [80, 100]  # height, width
+    top_left: [80, 100]  # height, width
     height: 200
   downsample: 1
 

configs/train_digicam_multimask.yaml

@@ -25,13 +25,13 @@ files:
       display_res: [900, 1200]  # used during measurement
       rotate: True   # if measurement is upside-down
       alignment:
-        topright: [80, 100]  # height, width
+        top_left: [80, 100]  # height, width
         height: 200
 
 # TODO: these parameters should be in the dataset?
 alignment:
   # when there is no downsampling
-  topright: [80, 100]  # height, width
+  top_left: [80, 100]  # height, width
   height: 200
 
 training:

configs/train_digicam_singlemask.yaml

@@ -24,13 +24,13 @@ files:
       display_res: [900, 1200]  # used during measurement
       rotate: True   # if measurement is upside-down
       alignment:
-        topright: [80, 100]  # height, width
+        top_left: [80, 100]  # height, width
         height: 200
 
 # TODO: these parameters should be in the dataset?
 alignment:
   # when there is no downsampling
-  topright: [80, 100]  # height, width
+  top_left: [80, 100]  # height, width
   height: 200
 
 training:

configs/train_unrolledADMM.yaml

@@ -41,7 +41,7 @@ files:
   extra_eval: null  # dict of extra datasets to evaluate on
 
 alignment: null
-#   topright: null    # height, width
+#   top_left: null    # height, width
   # height: null
 
 torch: True

lensless/hardware/slm.py

@@ -240,11 +240,11 @@ def adafruit_sub2full(
 
     # pad to full pattern
     pattern = np.zeros((3, 128, 160), dtype=np.uint8)
-    topleft = [center[0] - controllable_shape[1] // 2, center[1] - controllable_shape[2] // 2]
+    top_left = [center[0] - controllable_shape[1] // 2, center[1] - controllable_shape[2] // 2]
     pattern[
         :,
-        topleft[0] : topleft[0] + controllable_shape[1],
-        topleft[1] : topleft[1] + controllable_shape[2],
+        top_left[0] : top_left[0] + controllable_shape[1],
+        top_left[1] : top_left[1] + controllable_shape[2],
     ] = subpattern_rgb.astype(np.uint8)
     return pattern
 

lensless/utils/dataset.py

@@ -1064,7 +1064,7 @@ def __init__(
             If `psf` not provided, the SLM to use for the PSF simulation, by default "adafruit".
         alignment : dict, optional
             Alignment parameters between lensless and lensed data.
-            If "topright", "height", and "width" are provided, the region-of-interest from the reconstruction of ``lensless`` is extracted and ``lensed`` is reshaped to match.
+            If "top_left", "height", and "width" are provided, the region-of-interest from the reconstruction of ``lensless`` is extracted and ``lensed`` is reshaped to match.
             If "crop" is provided, the region-of-interest is extracted from the simulated lensed image, namely a ``simulation`` configuration should be provided within ``alignment``.
         return_mask_label : bool, optional
             If multimask dataset, return the mask label (True) or the corresponding PSF (False).
@@ -1106,11 +1106,11 @@ def __init__(
         self.crop = None
         if alignment is not None:
             # preparing ground-truth in expected shape
-            if "topright" in alignment:
+            if "top_left" in alignment:
                 self.alignment = dict(alignment.copy())
-                self.alignment["topright"] = (
-                    int(self.alignment["topright"][0] / downsample),
-                    int(self.alignment["topright"][1] / downsample),
+                self.alignment["top_left"] = (
+                    int(self.alignment["top_left"][0] / downsample),
+                    int(self.alignment["top_left"][1] / downsample),
                 )
                 self.alignment["height"] = int(self.alignment["height"] / downsample)
 
@@ -1304,14 +1304,14 @@ def extract_roi(self, reconstruction, lensed=None, axis=(1, 2)):
         assert max(axis) < n_dim, "Axis should be within the dimensions of the reconstruction."
 
         if self.alignment is not None:
-            top_right = self.alignment["topright"]
+            top_left = self.alignment["top_left"]
             height = self.alignment["height"]
             width = self.alignment["width"]
 
             # extract according to axis
             index = [slice(None)] * n_dim
-            index[axis[0]] = slice(top_right[0], top_right[0] + height)
-            index[axis[1]] = slice(top_right[1], top_right[1] + width)
+            index[axis[0]] = slice(top_left[0], top_left[0] + height)
+            index[axis[1]] = slice(top_left[1], top_left[1] + width)
             reconstruction = reconstruction[tuple(index)]
 
             # rotate if necessary

scripts/recon/digicam_mirflickr.py

@@ -84,10 +84,10 @@ def apply_pretrained(config):
     if save:
         print(f"Saving images to {os.getcwd()}")
         alignment = test_set.alignment
-        top_right = alignment["topright"]
+        top_left = alignment["top_left"]
         height = alignment["height"]
         width = alignment["width"]
-        res_np = img[top_right[0] : top_right[0] + height, top_right[1] : top_right[1] + width]
+        res_np = img[top_left[0] : top_left[0] + height, top_left[1] : top_left[1] + width]
         lensed_np = lensed[0].cpu().numpy()
         save_image(lensed_np, f"original_idx{idx}.png")
         save_image(res_np, f"{model_name}_idx{idx}.png")

scripts/recon/train_learning_based.py

@@ -318,13 +318,7 @@ def train_learned(config):
                         res_np, lensed_np = test_set.extract_roi(
                             res_np, lensed=lensed_np, axis=(0, 1)
                         )
-                    # if alignment is not None:
-                    #     top_right = alignment["topright"]
-                    #     height = alignment["height"]
-                    #     width = alignment["width"]
-                    #     res_np = res_np[
-                    #         top_right[0] : top_right[0] + height, top_right[1] : top_right[1] + width
-                    #     ]
+
                     cropped = True
 
                 elif config.training.crop_preloss: