Datasets:
Follow
Intelligent Vision and Robotics Lab at UT Dallas 6
Jun 29 03:55 Document grayscale fisheye WebP decode contract Clarify that standard WebP decoders expose grayscale frames as RGB and that safe_load_gray / Pillow L conversion restores the verified original plane. 03:44 [WIP] Lossless storage migration (WebP / PNG-9 / NPY) (#2 ) - Add files using upload-large-folder tool (95c60b6842acbfa4b4fa422360d0f556a53ed3cf)
- Add files using upload-large-folder tool (9e9cea3e781a0a91f8705d140564b60ece4ffdec)
- Add files using upload-large-folder tool (c6340166e52d983abee09a50cb17f8855e895305)
- Add files using upload-large-folder tool (af249595478ca50063786287972e980a35d4f961)
- Add files using upload-large-folder tool (037bca35d7d0205855957de6588dc1bd69e5cfcc)
- Add files using upload-large-folder tool (58b9ecc4bf9cf7fa4d6cf09886499883c38ea8a7)
- Add files using upload-large-folder tool (267607b7cedf5545d8b1c2cd4bb367a67c35e7f3)
- Add files using upload-large-folder tool (3ae72a845b3362cdb5cf004a14aae5059f989996)
- Add files using upload-large-folder tool (4045873ad50a0891c5f1cd0ceb7dd21767686772)
- Add files using upload-large-folder tool (1e80b20fe92d065507822c167daba59e531688cb)
- Add files using upload-large-folder tool (d1232fd8b133bf8cf6b410170627ea5bf7fbd4f9)
- Add files using upload-large-folder tool (e161c18d9c… Jun 9 04:02 Restore missing shard scenes/scene064/2/labels/masks_aux/v1.tar 04:01 Restore missing shard scenes/scene064/1/labels/masks_aux/v1.tar 03:59 Restore missing shard scenes/scene064/0/labels/masks_aux/v1.tar 03:57 Restore missing shard scenes/scene063/2/labels/cam_pose/v1.tar 03:57 Restore missing shard scenes/scene063/2/labels/sam2_meta/v1.tar 03:57 Restore missing shard scenes/scene063/2/labels/masks_aux/v1.tar 03:56 Restore missing shard scenes/scene063/2/labels/masks/v1.tar 03:55 Restore missing shard scenes/scene063/2/fisheye.tar 03:55 Restore missing shard scenes/scene063/2/depth.tar 03:55 Restore missing shard scenes/scene063/2/rgb.tar 03:55 Restore missing shard scenes/scene063/1/labels/cam_pose/v1.tar 03:55 Restore missing shard scenes/scene063/1/labels/sam2_meta/v1.tar 03:55 Restore missing shard scenes/scene063/1/labels/masks_aux/v1.tar 03:54 Restore missing shard scenes/scene063/1/labels/masks/v1.tar 03:54 Restore missing shard scenes/scene063/1/fisheye.tar 03:54 Restore missing shard scenes/scene063/1/rgb.tar 03:54 Restore missing shard scenes/scene063/0/labels/cam_pose/v1.tar 03:54 Restore missing shard scenes/scene063/0/labels/sam2_meta/v1.tar 03:54 Restore missing shard scenes/scene063/0/labels/masks_aux/v1.tar 03:54 Restore missing shard scenes/scene063/0/labels/masks/v1.tar 03:54 Restore missing shard scenes/scene063/0/rgb.tar 03:54 Restore missing shard scenes/scene016/2/labels/cam_pose/v1.tar 03:54 Restore missing shard scenes/scene016/2/labels/masks/v1.tar 03:54 Restore missing shard scenes/scene016/2/depth.tar 03:54 Restore missing shard scenes/scene016/1/labels/cam_pose/v1.tar 03:54 Restore missing shard scenes/scene016/1/fisheye.tar 03:53 Restore missing shard scenes/scene016/1/depth.tar 03:53 Restore missing shard scenes/scene016/1/rgb.tar 03:53 Restore missing shard scenes/scene016/0/fisheye.tar 03:53 Restore missing shard scenes/scene016/0/depth.tar 03:53 Restore missing shard scenes/scene016/0/rgb.tar Jun 5 18:43 Bust aligned jumbotron cache 18:43 Align jumbotron masks with source frames 18:34 Bust improved jumbotron palette cache 18:34 Improve jumbotron mask palette 18:28 Bust identity-colored jumbotron cache Updates the jumbotron video URL version query to request the identity-consistent mask-color MP4. 18:28 Use identity-consistent mask colors in jumbotron Rebuilds the jumbotron mask panels so object colors are assigned by object identity across phases instead of raw local mask IDs. 08:00 Bust jumbotron video cache Adds a version query to the dataset card jumbotron video source so clients fetch the rebuilt all-cells-synced MP4 immediately. 07:59 Sync all jumbotron cells from frame zero Rebuilds the jumbotron from raw phase data so every cell starts at source frame 0 together. Camera pose now reveals only the trajectory up to the current frame rather than showing future motion at the start. 07:48 Use absolute jumbotron video URL Use the Hub resolve URL for the dataset card video so the HTML video source resolves to the MP4 asset with video/mp4 MIME type. 07:44 Use MP4 jumbotron and remove release tooling Switches the dataset card video to browser-compatible MP4/H.264 and removes maintainer-only ego_release/rerun_check helper folders from the official dataset repo. Preview files remain because they back Dataset Viewer/Data Studio configs. 07:38 Update jumbotron and ego dataset metadata Adds the synced RPX jumbotron video, ego view manifests/previews, Dataset Viewer configs, and rerun/download helper updates. Jun 3 08:01 Clean Dataset Viewer preview typography and panel labels 07:22 Use embedded image media preview for Dataset Viewer 07:13 Expand Data Studio media preview to all MOS phases Replaces the 3-row media preview with one video-layout RGB/depth/mask still per MOS phase. Removes the old 3-file video preview folder to avoid confusion. Existing README teaser and jumbotron assets remain unchanged. 06:10 Use CSV-backed media preview for Data Studio Replaces binary media preview configs with a small CSV config containing direct preview image/video links, avoiding Dataset Viewer CSV parsing of binary files. Existing teaser and jumbotron assets remain unchanged. 06:05 Fix Data Studio preview config paths Use data_dir for image/video preview configs so Dataset Viewer treats them as media folders instead of CSV globs. Existing teaser/jumbotron assets remain unchanged. 05:58 Add Data Studio preview configs and ESD split tables Adds structured split CSVs plus small image/video preview configs for Hugging Face Data Studio. Keeps existing README teaser media and documents ESD split calculation. 
