2024-02-11 08:06:50 +00:00
|
|
|
# Formula image(grayscale) mean and variance
|
|
|
|
|
IMAGE_MEAN = 0.9545467
|
2025-02-28 19:56:49 +08:00
|
|
|
IMAGE_STD = 0.15394445
|
2024-02-11 08:06:50 +00:00
|
|
|
|
|
|
|
|
# Vocabulary size for TexTeller
|
2024-03-18 15:48:04 +00:00
|
|
|
VOCAB_SIZE = 15000
|
2024-02-11 08:06:50 +00:00
|
|
|
|
|
|
|
|
# Fixed size for input image for TexTeller
|
|
|
|
|
FIXED_IMG_SIZE = 448
|
|
|
|
|
|
|
|
|
|
# Image channel for TexTeller
|
|
|
|
|
IMG_CHANNELS = 1 # grayscale image
|
|
|
|
|
|
|
|
|
|
# Max size of token for embedding
|
2024-03-18 15:48:04 +00:00
|
|
|
MAX_TOKEN_SIZE = 1024
|
2024-02-11 08:06:50 +00:00
|
|
|
|
|
|
|
|
# Scaling ratio for random resizing when training
|
|
|
|
|
MAX_RESIZE_RATIO = 1.15
|
|
|
|
|
MIN_RESIZE_RATIO = 0.75
|
|
|
|
|
|
|
|
|
|
# Minimum height and width for input image for TexTeller
|
|
|
|
|
MIN_HEIGHT = 12
|
2025-02-28 19:56:49 +08:00
|
|
|
MIN_WIDTH = 30
|
2025-04-16 14:23:02 +00:00
|
|
|
|
|
|
|
|
LATEX_DET_MODEL_URL = (
|
|
|
|
|
"https://huggingface.co/TonyLee1256/texteller_det/resolve/main/rtdetr_r50vd_6x_coco.onnx"
|
|
|
|
|
)
|
|
|
|
|
TEXT_REC_MODEL_URL = (
|
|
|
|
|
"https://huggingface.co/OleehyO/paddleocrv4.onnx/resolve/main/ch_PP-OCRv4_server_rec.onnx"
|
|
|
|
|
)
|
|
|
|
|
TEXT_DET_MODEL_URL = (
|
|
|
|
|
"https://huggingface.co/OleehyO/paddleocrv4.onnx/resolve/main/ch_PP-OCRv4_det.onnx"
|
|
|
|
|
)
|