.PCD, .PLY, .LAS, .LAZ pointclouds

Overview

This option allows you to upload point clouds as episodes to the platform without any annotations. All items from the input directory and its subdirectories will be uploaded to a single dataset. If you need to preserve the directory structure, you can use the Import Pointcloud Episodes application from the Supervisely Ecosystem.

Format description

Supported image formats: .pcd. With annotations: No Supported annotation format: Not applicable. Grouped by: Any structure (will be uploaded to a single dataset).

Input files structure

Example data: download ⬇️ Example data with related images: download ⬇️

Recommended directory structure:

📦folder (with related images)          📦folder
 ┣ 📂pointcloud                          ┣ 📂pointcloud
 ┃ ┣ 📦0000000000.pcd                    ┃ ┣ 📦0000000000.pcd
 ┃ ┣ 📦0000000001.pcd                    ┃ ┣ 📦0000000001.pcd
 ┃ ┗ 📦0000000002.pcd                    ┃ ┗ 📦0000000002.pcd
 ┣ 📂related_images                      ┗ 📜frame_pointcloud_map.json
 ┃ ┣ 📂0000000000_pcd
 ┃ ┃ ┣ 🖼️0000000000.png
 ┃ ┃ ┗ 📜0000000000.png.json
 ┃ ┣ 📂0000000001_pcd
 ┃ ┃ ┣ 🖼️0000000001.png
 ┃ ┃ ┗ 📜0000000001.png.json
 ┃ ┣ 📂0000000002_pcd
 ┃ ┃ ┣ 🖼️0000000002.png
 ┃ ┃ ┗ 📜0000000002.png.json
 ┗ 📜frame_pointcloud_map.json                   

Frames mapping file structure:

📜frame_pointcloud_map.json
{
    "0": "0000000000.pcd",
    "1": "0000000001.pcd",
    "2": "0000000002.pcd"
}

Last updated

Was this helpful?