We're trying to train a bib detector here.
data_rawis just the place where the images are downloaded and wait to be annotated.datahas annotated data.data/bib_detection/annotationshas a collection of images in the format given by VGG Image Annotator. These images are ready to be processed bycreate_tf_records.pyto build the TF examples.data/number_recognition/annotationscontains images of bibs. Some of the images were cropped using the data from ground truth for bib detection and others were cropped using the predictions of the bib detector. All annotations are kept in a single filebibs.csv.
modelscontains code fromhttps://github.com/tensorflow/models. More specifically, it contains theobject_detection&attention_ocrmodels.
The images are grouped in directories by event. Each directory contains a few
images and one .csv file which contains the annotations.
This structure needs to be kept strictly or the create_tf_records.py script
won't work correctly.
- If required on your environment, activate Tensor Flow:
source tensorflow/bin/activate. - Set up the model:
bash run/setup_model.sh. - Start the training job:
bash run/train.sh. - Start the evaluation job:
bash run/eval.sh.
TODO