Skip to content

5. Obtaining Plots

Jump to bottom
Agam Dwivedi edited this page May 25, 2020 · 4 revisions

Plots

1.) Number of words(in logarithmic scale) vs Word Length

words vs wordlength

2.) Distribution of word-averaged erroneous character rate (WA-ECR) as a function of length, for different models

Word-avg error rate as function of length figure

To obtain these plots, run:

conda activate hcr

python visualize/obtain_waecr_plot.py <test_file_path> <predicted_test_file_byaocr_path> <predicted_test_file_by_crnn_path> <actual_crnn_test_file_path>

python visualize/obtain_waecr_plot.py label_data/annot_realTest.txt model/attention-lstm/logs/test_pred.txt model/CRNN/logs/crnn_pred.txt model/CRNN/logs/crnn_gt.txt

Clone this wiki locally