Pascal VOC
Overview
The Pascal VOC (Visual Object Classes) format stands as one of the benchmarks established relatively early for object classification, segmentation and detection. It furnishes a standardized dataset for identifying object classes, utilizing an XML-based export format that enjoys widespread adoption in computer vision tasks. This converter converts Pascal VOC format to Supervisely format. Learn more how to prepare data in Pascal VOC format and how to import the original Pascal VOC dataset in the Import Pascal VOC app.
Format description
Supported image formats: .jpg
, .jpeg
, .mpo
, .bmp
, .png
, .webp
, .tiff
, .tif
, .jfif
, .avif
, .heic
, and .heif
With annotations: Yes
Supported annotation files extensions: .xml
(bounding boxes), .png
(segmentation masks)
Grouped by: Any structure (will be uploaded as a single dataset)
Input files structure
Example data: download ⬇️
Pascal VOC archive or directory must have the following structure:
📦custom_pascal.tar 📂custom_pascal_project_dir
┗ 📂VOCdevkit ┗ 📂VOCdevkit
┗ 📂VOC or VOC2012 ┗ 📂VOC or VOC2012
┣ 📂Annotations ┣ 📂Annotations
┣ 📂ImageSets ┣ 📂ImageSets
┃ ┣ 📂Mainn ┃ ┣ 📂Main
┃ ┗ 📂Segmentation ┃ ┗ 📂Segmentation
┣ 📂JPEGImage ┣ 📂JPEGImages
┣ 📂SegmentationClasss ┣ 📂SegmentationClass
┣ 📂SegmentationObject ┣ 📂SegmentationObject
┗ 📜colors.txt ┗ 📜colors.txt
colors.txt
file is custom, and not provided in the original Pascal VOC Dataset. File contains information about instance mask colors associated with classes in Pascal VOC format. This file is required by this app, if you are uploading custom dataset. Each line of colors.txt
file starts with class_name
and ends with RGB
values that represent class color.
colors.txt
example:
neutral 224 224 192
kiwi 255 0 0
lemon 81 198 170
Useful links
Last updated
Was this helpful?