Skip to main content

Table 2 List of supported annotation formats

From: CLoDSA: a tool for augmentation in classification, localization, detection, semantic segmentation and instance segmentation tasks

Data

Problem

Input format

Output format

2D Images

Classification

A folder for each class of image

A folder for each class of image

   

An HDF5 file [26]

   

A Keras generator [12]

 

Localization

Pascal VOC format [21]

Pascal VOC format

   

An HDF5 file

 

Detection

Pascal VOC format

Pascal VOC format

  

YOLO format [27]

YOLO format

 

Segmentation

A folder containing the images and their associated masks

A folder containing the images and their associated masks

   

An HDF5 file

   

A Keras generator

 

Instance segmentation

COCO format [23]

COCO format

  

JSON format from ImageJ

JSON format from ImageJ

Multi-dimensional Images

Video Classification

A folder for each class of video

A folder for each class of video

 

Video Detection

Youtube BB format [28]

Youtube BB format

 

Stack segmentation

Pairs of tiff files containing the stack and the associated mask

Pairs of tiff files containing the stack and the associated mask