Closed hounaidaMoallaF closed 4 years ago
See #40.
I train using a platform with multiple GPUS, and the graph in the training checkpoints reflects this; they are not suitable for general use. Hence, if someone wrote script that harnesses only the model parameters necessary for test, I would gladly use it to package a test model. I don't have time nor need for such a script myself.
thank you verry mutch for your response
On Sun, Nov 25, 2018 at 1:49 PM Jerod Weinman notifications@github.com wrote:
See #40 https://github.com/weinman/cnn_lstm_ctc_ocr/issues/40.
I train using a platform with multiple GPUS, and the graph in the training checkpoints reflects this; they are not suitable for general use. Hence, if someone wrote script that harnesses only the model parameters necessary for test, I would gladly use it to package a test model. I don't have time nor need for such a script myself.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/weinman/cnn_lstm_ctc_ocr/issues/45#issuecomment-441441788, or mute the thread https://github.com/notifications/unsubscribe-auth/AkI1-fbLD3ykAsnuespeP-DTOC2ns2jwks5uyp_vgaJpZM4Yxkg5 .
The model checkpoint (trained using MapTextSynth) from the ICDAR 2019 paper is posted here: https://weinman.cs.grinnell.edu/data/cnn_lstm_ctc_ocr_checkpoint_icdar19.tar.bz2
Please, can you give me you pre-trained model for test