lzy/sci-gui-agent-benchmark
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Support downsampling; Fix bugs in windows a11y tree; Add a11y_tree trim

Browse Source
This commit is contained in:
Timothyxxx
2024-03-25 18:02:48 +08:00
parent 635b6717b3
commit 172123ab2c
4 changed files with 104 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
mm_agents/accessibility_tree_wrap/heuristic_retrieve.py
View File

@@ -110,6 +110,10 @@ def draw_bounding_boxes(nodes, image_file_path, output_image_file_path, down_sam
coords = tuple(map(int, coords_str.strip('()').split(', ')))
size = tuple(map(int, size_str.strip('()').split(', ')))
import copy
original_coords = copy.deepcopy(coords)
original_size = copy.deepcopy(size)
if float(down_sampling_ratio) != 1.0:
# Downsample the coordinates and size
coords = tuple(int(coord * down_sampling_ratio) for coord in coords)
@@ -145,7 +149,7 @@ def draw_bounding_boxes(nodes, image_file_path, output_image_file_path, down_sam
draw.text(text_position, str(index), font=font, anchor="lb", fill="white")
# each mark is an x, y, w, h tuple
marks.append([coords[0], coords[1], size[0], size[1]])
marks.append([original_coords[0], original_coords[1], original_size[0], original_size[1]])
drew_nodes.append(_node)
if _node.text: