1LSTMEVAL(1)                                                        LSTMEVAL(1)
2
3
4

NAME

6       lstmeval - Evaluation program for LSTM-based networks.
7

SYNOPSIS

9       lstmeval --model
10       lang.lstm|modelname_checkpoint|modelname_N.NN_NN_NN.checkpoint
11       [--traineddata lang/lang.traineddata] --eval_listfile
12       lang.eval_files.txt [--verbosity N] [--max_image_MB NNNN]
13

DESCRIPTION

15       lstmeval(1) evaluates LSTM-based networks. Either a recognition model
16       or a training checkpoint can be given as input for evaluation along
17       with a list of lstmf files. If evaluating a training checkpoint,
18       --traineddata should also be specified. Intermediate training
19       checkpoints can also be used.
20

OPTIONS

22       --model FILE
23           Name of model file (training or recognition) (type:string default:)
24
25       --traineddata FILE
26           If model is a training checkpoint, then traineddata must be the
27           traineddata file that was given to the trainer (type:string
28           default:)
29
30       --eval_listfile FILE
31           File listing sample files in lstmf training format. (type:string
32           default:)
33
34       --max_image_MB INT
35           Max memory to use for images. (type:int default:2000)
36
37       --verbosity INT
38           Amount of diagnosting information to output (0-2). (type:int
39           default:1)
40

HISTORY

42       lstmeval(1) was first made available for tesseract4.00.00alpha.
43

RESOURCES

45       Main web site: https://github.com/tesseract-ocr Information on training
46       tesseract LSTM:
47       https://tesseract-ocr.github.io/tessdoc/TrainingTesseract-4.00.html
48

SEE ALSO

50       tesseract(1)
51

COPYING

53       Copyright (C) 2012 Google, Inc. Licensed under the Apache License,
54       Version 2.0
55

AUTHOR

57       The Tesseract OCR engine was written by Ray Smith and his research
58       groups at Hewlett Packard (1985-1995) and Google (2006-present).
59
60
61
62                                  02/25/2022                       LSTMEVAL(1)
Impressum