OCR: Add IAM corpus with unk decoding support #6
Conversation
|
Ashish, could you please rebase against the ocr branch? |
|
|
||
| num_targets=$(tree-info $tree_dir/tree | grep num-pdfs | awk '{print $2}') | ||
| learning_rate_factor=$(echo "print 0.5/$xent_regularize" | python) | ||
| common1="required-time-offsets= height-offsets=-2,-1,0,1,2 num-filters-out=36" |
There was a problem hiding this comment.
you can remove required-time-offsets= altogether
aarora8
force-pushed
the
iam
branch
2 times, most recently
from
6a93702 to
f8eb4fd
Compare
|
Thanks, rebased it against ocr branch. updated headers in the new recipes. |
| @@ -29,8 +29,8 @@ alignment_subsampling_factor=1 | |||
| chunk_width=340,300,200,100 | |||
There was a problem hiding this comment.
Please update the results for this recipe if it's not already updated
| @@ -33,8 +33,8 @@ alignment_subsampling_factor=1 | |||
| chunk_width=340,300,200,100 | |||
There was a problem hiding this comment.
Also for this recipe.
Change the description to "chainali_1a is as 1a except it uses chain alignments (using 1a system) instead of gmm alignments" and then append the output (and the command itself) of compare_wer.sh for 1a and chainali_1a (after 1 blank line)
| @@ -0,0 +1,235 @@ | |||
| #!/bin/bash | |||
|
|
|||
| # chainali_1b uses chain model for lattice instead of gmm-hmm model. It has more cnn layers as compared to 1a | |||
There was a problem hiding this comment.
change this to "chainali_1b is as chainali_1a except it has 3 more cnn layers."
Then append the compare_wer.sh output (with the command) after adding a blank line
| @@ -0,0 +1,226 @@ | |||
| #!/bin/bash | |||
There was a problem hiding this comment.
please remove this and 1d for now. The improvements are not significant.
|
sorry. updated headers for run_cnn_1a.sh, run_cnn_chainali_1a.sh, run_cnn_chainali_1b.sh. removed run_cnn_chainali_1c.sh , run_cnn_chainali_1d.sh. |
|
Thanks. Merging... |
* OCR: Add IAM corpus with unk decoding support (#3) * Add a new English OCR database 'UW3' * Some minor fixes re IAM corpus * Fix an issue in IAM chain recipes + add a new recipe (#6) * Some fixes based on the pull request review * Various fixes + cleaning on IAM * Fix LM estimation and add extended dictionary + other minor fixes * Add README for IAM * Add output filter for scoring * Fix a bug RE switch to pyhton3 * Add updated results + minor fixes * Remove unk decoding -- gives almost no gain * Add UW3 OCR database * Fix cmd.sh in IAM + fix usages of train/decode_cmd in chain recipes * Various minor fixes on UW3 * Rename iam/s5 to iam/v1 * Add README file for UW3 * Various cosmetic fixes on UW3 scripts * Minor fixes in IAM
* OCR: Add IAM corpus with unk decoding support (#3) * Add a new English OCR database 'UW3' * Some minor fixes re IAM corpus * Fix an issue in IAM chain recipes + add a new recipe (#6) * Some fixes based on the pull request review * Various fixes + cleaning on IAM * Fix LM estimation and add extended dictionary + other minor fixes * Add README for IAM * Add output filter for scoring * Fix a bug RE switch to pyhton3 * Add updated results + minor fixes * Remove unk decoding -- gives almost no gain * Add UW3 OCR database * Fix cmd.sh in IAM + fix usages of train/decode_cmd in chain recipes * Various minor fixes on UW3 * Rename iam/s5 to iam/v1 * Add README file for UW3 * Various cosmetic fixes on UW3 scripts * Minor fixes in IAM
No description provided.