Datasets for ichthyolith detection

The following three kinds of dataset are shared.

(1) 01_dataset_for_ai_ichthyolith
A zip file of training and validation dataset for ai_ichthyolith (Mask R-CNN).
A training folder includes 866 images and 1 annotation information in the format of VGG Image Annotator (https://www.robots.ox.ac.uk/~vgg/software/via/).
A validation folder include 92 images and 1 annotation information.

(2) 02_dataset_for_efficientNet-v2
A zip file for training and validation of classification model (efficientnet-v2).
A dataset include 17400 images of tooth, 15036 images of noise.
These images were randomly split into training and validation dataset in the training code.
Although 36 images of denticle are included, they were not used for classification, because the number is too small.

(3) 03_dataset_for_practical_test
6 Zip files that include whole the observation regions of the slide.
A file include ~1000 images and annotation information in the format of labelImg (https://github.com/tzutalin/labelImg).

Sample codes are available at https://github.com/KazuhideMimura/ai_ichthyolith and https://github.com/KazuhideMimura/eNetV2_for_ai_ichthyolith