Dataset Viewer
Auto-converted to Parquet Duplicate
scene_id
stringclasses
9 values
display_name
stringclasses
9 values
archive
stringclasses
9 values
archive_sha256
stringclasses
9 values
archive_bytes
int64
129M
358M
files_in_zip
int32
811
1.93k
traversals
int8
3
3
sampling_stride
int8
5
10
frames
int32
798
1.91k
annotated_frames
int32
262
407
annotation_points
int32
398
612
labels
listlengths
10
10
coordinate_frame
stringclasses
1 value
contains_las
bool
1 class
2metro8floor
2 MetroTech, 8th floor
archives/dejaview_v2_2metro8floor.zip
f36f7b6a40e2fabe33d50ee674a4e2fa12bea3070402266946eca3aa8d7c9e28
145,263,961
941
3
10
928
262
430
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
6metro6floor
6 MetroTech, 6th floor
archives/dejaview_v2_6metro6floor.zip
49cfd17b6eb33fe367c5a3ebc56e56af50a50d4e5a9db46434e80ed6dd763ce5
136,137,211
819
3
10
806
278
434
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
370jayst2floor
370 Jay Street, 2nd floor
archives/dejaview_v2_370jayst2floor.zip
fab75336173ebb6a9b4fa5b2051590f154f04dc1cf2ad1c488794b8bc1b5d148
159,311,204
875
3
10
862
262
472
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
370jayst3floor
370 Jay Street, 3rd floor
archives/dejaview_v2_370jayst3floor.zip
0f5169f9f5ae6235bee3120601bed39107f33c908c74624c418ee17ea01ecd84
149,116,637
913
3
10
900
275
398
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
370jayst5floor
370 Jay Street, 5th floor
archives/dejaview_v2_370jayst5floor.zip
54afc4039b6912c0cc2ca473f57db83fe611fd23002a7d3bfb94ffadb981af73
128,637,297
811
3
10
798
267
412
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
370jayst8floor
370 Jay Street, 8th floor
archives/dejaview_v2_370jayst8floor.zip
17db0e39c3b2240a982f216a3b6b755de0070035938e034eedbd38623e4d8b8e
328,020,450
1,927
3
5
1,914
353
443
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
greyson3floor
Greyson, 3rd floor
archives/dejaview_v2_greyson3floor.zip
6a039d579579e0c523aa03cbbc46c58095fb2bc3cd1514dab5f11914d5ac433c
244,302,995
1,460
3
5
1,446
407
581
[ "Paper_Plate", "Book", "Pencil", "Blue_Tape", "Green_Tape", "Scissors", "Glove", "Water_Bottle", "Cheez_It", "Remote_Control" ]
run_1
false
columbus90indoor
Columbus 90, indoor
archives/dejaview_v2_columbus90indoor.zip
6df2ba7a8ae1128da4e22a70d1f52236d698c39b20307323b22bcb405e8108a9
163,236,636
884
3
10
870
357
538
[ "Calculator", "Remote_Control", "Umbrella", "Green_Tape", "Headphones", "Yellow_Hat", "Yellow_Coat", "Book", "Water_Bottle", "Glove" ]
run_1
false
columbus90outdoor
Columbus 90, outdoor
archives/dejaview_v2_columbus90outdoor.zip
353785c06fda6dab29d5073b32dc8563cb5d356059dbdce10572b88bef5c6264
357,656,988
1,182
3
10
1,168
350
612
[ "Calculator", "Remote_Control", "Umbrella", "Green_Tape", "Headphones", "Yellow_Hat", "Yellow_Coat", "Book", "Water_Bottle", "Glove" ]
run_1
false

DejaView v2

DejaView v2 is a clean three-traversal indoor/outdoor dataset distributed as one ZIP per physical scene. It contains 9 scenes and 27 aligned traversals.

Scene Stride Frames Annotated frames Points ZIP size
2metro8floor 10 928 262 430 145.3 MB
6metro6floor 10 806 278 434 136.1 MB
370jayst2floor 10 862 262 472 159.3 MB
370jayst3floor 10 900 275 398 149.1 MB
370jayst5floor 10 798 267 412 128.6 MB
370jayst8floor 5 1914 353 443 328.0 MB
greyson3floor 5 1446 407 581 244.3 MB
columbus90indoor 10 870 357 538 163.2 MB
columbus90outdoor 10 1168 350 612 357.7 MB

Distribution

data/scenes.parquet contains one row per scene and points to the corresponding file under archives/. The train split means all released scenes; it is not a prescribed machine-learning train split.

Each ZIP is self-contained:

DATASET_CARD.md
SCENE_SHA256SUMS.txt
<scene_id>/
  images/run_N_<timestamp_ns>_<left|right>.jpg
  stream_annotations.json
  camera_poses.json
  labels.json
  scene_manifest.json
  alignment.json
  runs/run_N/odom.csv
  runs/run_N/ImgPose.txt

Images are 800x800 undistorted pinhole views with 90-degree field of view and fx=fy=cx=cy=400 pixels. camera_to_world, dense ImgPose.txt, and odometry for all three runs are expressed in the run-1 coordinate frame.

The hardware records the right image first. Exactly the first right frame of each run has new_traversal: true; the paired left frame is false.

Point annotations use 15 categories: Paper_Plate, Book, Pencil, Blue_Tape, Green_Tape, Scissors, Glove, Water_Bottle, Cheez_It, Remote_Control, Calculator, Umbrella, Headphones, Yellow_Hat, Yellow_Coat.

Deliberate exclusions

Colorized LAS point clouds are not included. They were used only to visually check boundary alignment and are not inputs to the released task. The source captures, DA3/VGGT outputs, embeddings, semantic voxel maps, evaluation results, logs and processing intermediates are also excluded.

SHA256SUMS.txt, upload_manifest.json, and each ZIP's SCENE_SHA256SUMS.txt provide integrity metadata. No public data license has been assigned; keep the repository private until the owner selects a license.

Timestamp audits

Greyson run 2 is normalized by +180 seconds to remove its overlap with run 1; the scene ZIP contains the reversible mapping. The two Columbus 90 scenes were checked pairwise across all three runs and have no timestamp overlap, so their offsets are zero. Each Columbus ZIP contains timestamp_audit.json.

Downloads last month
69