sebastien
/
Copies
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Gemini for labels

master
Sébastien Miquel 2026-02-08 17:41:04 +01:00
parent 7575f61dbd
commit e79059ef64
1 changed files with 29 additions and 19 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
gemini_dir_batching.py → gemini_for_labels.py
View File

@ -167,30 +167,40 @@ def generate_request(file, labels, names, context_labels):
return (contents, generate_content_config) return (contents, generate_content_config)
# Argument Parsing # Argument Parsing
parser = argparse.ArgumentParser(description="Process a directory or specific file using Gemini.") parser = argparse.ArgumentParser(description="Process a directory or specific files using Gemini.")
parser.add_argument("input_path", help="The input directory or specific file") parser.add_argument("input_paths", nargs='+', help="The input directory or specific files")
parser.add_argument("--overwrite", action="store_true", help="Regenerate output even if it exists") parser.add_argument("--overwrite", action="store_true", help="Regenerate output even if it exists")
args = parser.parse_args() args = parser.parse_args()
input_arg = Path(args.input_path) # input_arg = Path(args.input_path)
image_files = [] image_files = []
# Setup Paths and Files def natural_key(text):
if input_arg.is_file(): return [int(c) if c.isdigit() else c.lower() for c in re.split(r'(\d+)', str(text))]
INPUT_DIR = input_arg.parent
CUTLEFT_DIR = INPUT_DIR / 'Cutleft' for path_str in args.input_paths:
# For a single file, we verify it exists but we might miss context if we don't look for siblings input_arg = Path(path_str)
# Simplification: We add just this file, context will be empty.
target_image = CUTLEFT_DIR / f"{input_arg.stem}.jpg" if input_arg.is_file():
if target_image.exists(): INPUT_DIR = input_arg.parent
image_files = [target_image] CUTLEFT_DIR = INPUT_DIR / 'Cutleft'
# Matches stem_01.jpg, stem_02.jpg, etc.
found_files = sorted(list(CUTLEFT_DIR.glob(f"{input_arg.stem}_*.jpg")),
key=natural_key)
if found_files:
image_files.extend(found_files)
else:
print(f"Warning: No variants found for {input_arg.stem} in {CUTLEFT_DIR}")
elif input_arg.is_dir():
INPUT_DIR = input_arg
CUTLEFT_DIR = INPUT_DIR / 'Cutleft'
image_files.extend(sorted(list(CUTLEFT_DIR.glob("*.jpg")), key=natural_key))
else: else:
print(f"Error: {target_image} not found.") print(f"Error: {input_arg} is not a valid file or directory.")
sys.exit(1)
else:
INPUT_DIR = input_arg
CUTLEFT_DIR = INPUT_DIR / 'Cutleft'
image_files = sorted(list(CUTLEFT_DIR.glob("*.jpg")))
labels_txt = (INPUT_DIR / "labels").read_text() labels_txt = (INPUT_DIR / "labels").read_text()
names_txt = (INPUT_DIR / "names").read_text() names_txt = (INPUT_DIR / "names").read_text()
@ -240,7 +250,7 @@ def process_copy_group(group_key, files):
pass # If read fails, next part has no context pass # If read fails, next part has no context
continue continue
print(f"[{group_key}] Processing {image_file.name} with {len(accumulated_labels)} ctx items...") print(f"[{group_key}] Processing {image_file.name} with {len(accumulated_labels)} accumulated labels...")
try: try:
contents, config = generate_request(image_file, labels_txt, names_txt, accumulated_labels) contents, config = generate_request(image_file, labels_txt, names_txt, accumulated_labels)