Skip to content

Latest commit

 

History

History
17 lines (11 loc) · 1 KB

readme.md

File metadata and controls

17 lines (11 loc) · 1 KB

EAST: An Efficient and Accurate Scene Text Detector - see readme.md.org

Here is to introduce for how to generate the frozen model file from the pretrained model

Download

Models trained on ICDAR 2013 (training set) + ICDAR 2015 (training set): BaiduYun link GoogleDrive

Run

# python3 export_model.py east_icdar2015_resnet_v1_50_rbox/model.ckpt-49491.meta           east_icdar2015_resnet_v1_50_rbox/model.ckpt-49491
# python3 ../tf_src/tensorflow/tensorflow/python/tools/freeze_graph.py --input_graph=./output_model/pb_model/model.pb --input_checkpoint east_icdar2015_resnet_v1_50_rbox/model.ckpt-49491 --output_graph=./output_model/pb_model/model_frozen.pb --output_node_names=feature_fusion/concat_3
# tensorboard --logdir ./logs/

The feature_fusion/concat_3 is the last layer which is defined in model.py. It can also be found by using tensorboard tool

Tensorboard