README.md

LAS-OCR

OCR model made of using LAS's speller and image encoder.

---

Example

from model import InferenceModel

model_path = './models/LAS_number_enclstm3_declstm3_3440.pt'
inference_model = InferenceModel(model_path)

import matplotlib.pyplot as plt
import cv2

img = cv2.imread(your_img_path)
s, _ = inference_model(img, is_bgr=True, use_beam=True)
print(s)

plt.imshow(img[...,::-1])

51.55

---

Preview

Loss Graph

---

Visual Attention Score

---

Reference

Listen-Attend-and-Spell-Pytorch

Chan, William, et al. “Listen, attend and spell.” arXiv preprint arXiv:1508.01211 (2015).APA