Skip to content

Shun14/TextBoxes_plusplus_Tensorflow

Repository files navigation

TextBoxes++-TensorFlow

TextBoxes++ re-implementation using tensorflow. This project is greatly inspired by slim project And many functions are modified based on SSD-tensorflow project

Author: Zhisheng Zou [email protected]

pretrained model

  1. Google drive

environment

python2.7/python3.5

tensorflow-gpu 1.8.0

at least one gpu

how to use

  1. Getting the xml file like this example xml and put the image together because we need the format like this standard xml
    1. picture format: *.png or *.PNG
  2. Getting the xml and flags ensure the XML file is under the same directory as the corresponding image.execute the code: convert_xml_format.py
    1. python tools/convert_xml_format.py -i in_dir -s split_flag -l save_logs -o output_dir
    2. in_dir means the absolute directory which contains the pic and xml
    3. split_flag means whether or not to split the datasets
    4. save_logs means whether to save train_xml.txt
    5. output_dir means where to save xmls
  3. Getting the tfrecords
    1. python gene_tfrecords.py --xml_img_txt_path=./logs/train_xml.txt --output_dir=tfrecords
    2. xml_img_txt_path like this train xml
    3. output_dir means where to save tfrecords
  4. Training
    1. python train.py --train_dir =some_path --dataset_dir=some_path --checkpoint_path=some_path
    2. train_dir store the checkpoints when training
    3. dataset_dir store the tfrecords for training
    4. checkpoint_path store the model which needs to be fine tuned
  5. Testing
    1. python test.py -m /home/model.ckpt-858 -o test
    2. -m which means the model
    3. -o which means output_result_dir
    4. -i which means the test img dir
    5. -c which means use which device to run the test
    6. -n which means the nms threshold
    7. -s which means the score threshold

Note:

  1. when you are training the model, you can run the eval_result.py to eval your model and save the result

About

Textboxes_plusplus implementation with Tensorflow (python)

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages