deep-text-recognition-benchmark

deep-text-recognition-benchmark copied to clipboard

learning rate scheduler

2

hi thanks for your code and best paper I do not see learning rate scheduler in training loop,it don not affect in performance of model?

raminrahimi6970

training dataset problem

5

Hi, I would like to make the model working well on both printed and handwritten papers. I am confused about the way to train the model. So, if I want...

hannah4kr

With MJ and ST the batch ratio is 0.5 0.5 respectively. Can someone guide me, what would be the best approach when adding the special character data into the training...

SaeedArisha

Special Characters Recognition

2

Hi, I am trying t use the ST_SPE.zip lmdb dataset of the special characters to fine-tune another model which has been previously trained for non-special characters but the model does...

spoorgholi74

lmdb.Error: training/.ipynb_checkpoints: No such file or directory

6

!python train.py --train_data='training' --valid_data='training' --Transformation='TPS' --FeatureExtraction='ResNet' --SequenceModeling='BiLSTM' --Prediction='Attn' ---------------------------------------------------------------------------------------------------------------------------------------- Filtering the images containing characters which are not in opt.character Filtering the images whose label is longer than opt.batch_max_length ['1'] dataset_root:...

Tejas-Tripathi

Fine tuning problem! urgent

4

So i have trained from scratch using lmdb dataset from the repo then i tried to create my own dataset using trdg to add non latin characters and retrain the...

gitdeepheolp

TypeError: __init__() got an unexpected keyword argument 'num_class'

4

Hi, please explain me what this error may be related to. I do everything according to the instructions, I made three files to reader, but in the *.yaml file I...

lenanazarova

Help training my dataset

2

So, i started creating my learning images using trdg , moved data.mdb lock.mdb to /data_lmdb_release/training/ and data_lmdb_release/validation used the command to train : CUDA_VISIBLE_DEVICES=0 python3.8 train.py --train_data data_lmdb_release/training --valid_data data_lmdb_release/validation...

gitdeepheolp

Is there any tip on super multiclass classifying?

1

Could you provide some suggestions on the classification of super multiple categories (such as 3000 categories)? An attempt was made to use a 3000-word dictionary, but the model did not...

xyh01

ValueError: num_samples should be a positive integer value, but got num_samples=0

3

train.py --train_data lmdb_dataset/training --valid_data lmdb_dataset/valid --Transformation TPS --FeatureExtraction ResNet --SequenceModeling BiLSTM --Prediction Attn Filtering the images containing characters which are not in opt.character Filtering the images whose label is longer...

Amithvassit