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

Fix one multi_app example; remove some broken examples; Support downsampling

Browse Source
This commit is contained in:
Timothyxxx
2024-03-21 22:05:16 +08:00
parent 7ca91ca8c9
commit 3ce7636abd
5 changed files with 11 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
mm_agents/accessibility_tree_wrap/heuristic_retrieve.py
View File

@@ -80,9 +80,11 @@ def filter_nodes(root: ET, platform="ubuntu", check_image=False):
return filtered_nodes
def draw_bounding_boxes(nodes, image_file_path, output_image_file_path):
def draw_bounding_boxes(nodes, image_file_path, output_image_file_path, down_sampling_ratio=1.0):
# Load the screenshot image
image = Image.open(image_file_path)
if float(down_sampling_ratio) != 1.0:
image = image.resize((int(image.size[0] * down_sampling_ratio), int(image.size[1] * down_sampling_ratio)))
draw = ImageDraw.Draw(image)
marks = []
drew_nodes = []
@@ -108,6 +110,11 @@ def draw_bounding_boxes(nodes, image_file_path, output_image_file_path):
coords = tuple(map(int, coords_str.strip('()').split(', ')))
size = tuple(map(int, size_str.strip('()').split(', ')))
if float(down_sampling_ratio) != 1.0:
# Downsample the coordinates and size
coords = tuple(int(coord * down_sampling_ratio) for coord in coords)
size = tuple(int(s * down_sampling_ratio) for s in size)
# Check for negative sizes
if size[0] <= 0 or size[1] <= 0:
raise ValueError(f"Size must be positive, got: {size}")