Quickstart Trajectories#

A small video dataset for trajectory annotation.

The dataset consists of the same 10 video segments as quickstart-video, but the per-frame object detections are linked across frames into object trajectories via the Instance values in their instance attributes (the legacy index attribute has been removed).

Note

Before annotating trajectories, you must compute video metadata and sample the videos into frames:

dataset.compute_metadata()
dataset.to_frames(sample_frames=True)

Details

  • Dataset name: quickstart-trajectories

  • Dataset size: 35.20 MB

  • Dataset license: CC-BY-4.0

  • Tags: video, quickstart

  • Supported splits: N/A

  • ZooDataset class: QuickstartTrajectoriesDataset

Example usage

1import fiftyone as fo
2import fiftyone.zoo as foz
3
4dataset = foz.load_zoo_dataset("quickstart-trajectories")
5
6session = fo.launch_app(dataset)

Note

In order to work with video datasets, you’ll need to have ffmpeg installed.

quickstart-trajectories