`sparkocr.eval_hunyuan_funsd`#

Module Contents#

Functions#

`clean_repeated_substrings`(text)
`parse_hunyuan_output`(text)	Convert:
`process_image`(img_path)

Attributes#

`FUNSD_ROOT`
`MAX_NEW_TOKENS`
`MODEL_NAME`
`OUTPUT_DIR`
`model`
`processor`
`split_dir`

clean_repeated_substrings(text)#

parse_hunyuan_output(text)#

Convert:: “Hello world(10,20),(50,60)Another text(70,80),(120,140)”
into:: [{“text”: “…”, “coordinates”: [x1,y1,x2,y2]}, …]

process_image(img_path)#

FUNSD_ROOT = './dataset/images'#

MAX_NEW_TOKENS = 16384#

MODEL_NAME = 'tencent/HunyuanOCR'#

OUTPUT_DIR = './funsd_hunyuan_outputs'#

model#

processor#

split_dir#