Dataset Viewer
Auto-converted to Parquet Duplicate
frame_id
stringlengths
6
6
points
listlengths
59.4k
124k
num_points
int64
14.9k
31k
point_dim
int64
4
4
spoof_gt
listlengths
7
7
gt_names
listlengths
1
21
gt_boxes
listlengths
7
147
num_gt
int64
1
21
gt_difficulty
listlengths
1
21
gt_num_points
listlengths
1
21
calib_P2
listlengths
16
16
calib_R0_rect
listlengths
16
16
calib_Tr_velo_to_cam
listlengths
16
16
image_shape
listlengths
2
2
atk_perturbation
stringclasses
1 value
atk_seed
int64
42
3.81k
atk_offset
listlengths
2
2
atk_yaw
float64
-0.3
0.3
atk_n_spoof_r
int64
225
12.3k
atk_n_spoof_k
int64
6
12.3k
000001
[49.52000045776367,22.667999267578125,2.0510001182556152,0.0,49.428001403808594,22.81399917602539,2.(...TRUNCATED)
18,862
4
[19.899812698364258,4.30317497253418,-0.851006031036377,4.429999828338623,1.5199999809265137,1.47000(...TRUNCATED)
[ "Truck", "Car", "Cyclist" ]
[69.72479248046875,-0.4475646913051605,0.5836524367332458,12.34000015258789,2.630000114440918,2.8499(...TRUNCATED)
3
[ 1, -1, -1 ]
[ 71, 9, 18 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
42
[ 19.899812698364258, 4.30317497253418 ]
0.118421
360
351
000002
[70.17500305175781,2.3550000190734863,2.5829999446868896,0.0,69.98300170898438,2.569000005722046,2.5(...TRUNCATED)
20,128
4
[12.787956237792969,-5.759644985198975,-1.0509142875671387,3.9200000762939453,1.6399999856948853,1.3(...TRUNCATED)
[ "Misc", "Car" ]
[8.83980941772461,-3.2139267921447754,-0.7918716073036194,2.369999885559082,1.4800000190734863,1.629(...TRUNCATED)
2
[ 0, 1 ]
[ 1349, 67 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
43
[ 12.787956237792969, -5.759644985198975 ]
0.203528
460
82
000004
[57.65700149536133,20.996000289916992,2.2839999198913574,0.0,57.60200119018555,21.180999755859375,2.(...TRUNCATED)
19,195
4
[16.64603614807129,-1.1307512521743774,-0.7613164186477661,4.710000038146973,1.7400000095367432,1.73(...TRUNCATED)
[ "Car", "Car" ]
[38.5496940612793,15.734733581542969,-0.921228289604187,4.010000228881836,1.7599999904632568,1.49000(...TRUNCATED)
2
[ 1, -1 ]
[ 78, 26 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
44
[ 16.64603614807129, -1.1307512521743774 ]
0.28151
532
514
000005
[51.81399917602539,11.093999862670898,2.000999927520752,0.25,51.81399917602539,11.265000343322754,2.(...TRUNCATED)
20,279
4
[21.512840270996094,3.1638028621673584,-0.9237403273582458,3.490000009536743,1.559999942779541,1.330(...TRUNCATED)
[ "Pedestrian" ]
[23.311281204223633,8.522290229797363,-0.8766786456108093,0.6499999761581421,0.9599999785423279,1.87(...TRUNCATED)
1
[ 0 ]
[ 70 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
45
[ 21.512840270996094, 3.1638028621673584 ]
0.187016
556
547
000006
[64.59400177001953,12.885000228881836,2.436000108718872,0.10000000149011612,64.63400268554688,13.104(...TRUNCATED)
19,751
4
[13.390082359313965,-2.7291629314422607,-0.8045405149459839,4.110000133514404,1.6299999952316284,1.4(...TRUNCATED)
[ "Car", "Car", "Car", "Car" ]
[48.47077178955078,2.959061622619629,-0.25629204511642456,3.619999885559082,1.559999942779541,1.4800(...TRUNCATED)
4
[ -1, 0, 0, 1 ]
[ 9, 64, 321, 26 ]
[718.3350830078125,0.0,600.3890991210938,44.50381851196289,0.0,718.3350830078125,181.51220703125,-0.(...TRUNCATED)
[0.9999477863311768,0.009791706688702106,-0.002925304928794503,0.0,-0.009806939400732517,0.999938189(...TRUNCATED)
[0.007755449041724205,-0.9999694228172302,-0.0010143029503524303,-0.007275538053363562,0.00229405588(...TRUNCATED)
[ 374, 1238 ]
injection
46
[ 13.390082359313965, -2.7291629314422607 ]
0.07311
820
659
000008
[21.554000854492188,0.02800000086426735,0.9380000233650208,0.3400000035762787,21.239999771118164,0.0(...TRUNCATED)
17,344
4
[25.566038131713867,-0.417833536863327,-0.9098907709121704,2.5399999618530273,1.5399999618530273,1.4(...TRUNCATED)
[ "Car", "Car", "Car", "Car", "Car", "Car" ]
[3.9702506065368652,2.716721534729004,-0.9451114535331726,3.2300000190734863,1.5700000524520874,1.60(...TRUNCATED)
6
[ -1, 1, -1, 1, 0, 0 ]
[ 1320, 1900, 878, 659, 55, 162 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
47
[ 25.566038131713867, -0.417833536863327 ]
-0.237765
1,397
252
000015
[49.54999923706055,0.09799999743700027,1.8849999904632568,0.17000000178813934,49.48099899291992,0.25(...TRUNCATED)
18,361
4
[22.723922729492188,0.1656065434217453,-0.8232946991920471,4.429999828338623,1.7300000190734863,1.63(...TRUNCATED)
[ "Car", "Pedestrian", "Pedestrian", "Pedestrian", "Pedestrian" ]
[4.348688125610352,2.783051013946533,-0.9599606394767761,4.139999866485596,1.6699999570846558,1.5700(...TRUNCATED)
5
[ -1, 1, 0, 0, 0 ]
[ 1645, 383, 55, 70, 66 ]
[718.3350830078125,0.0,600.3890991210938,44.50381851196289,0.0,718.3350830078125,181.51220703125,-0.(...TRUNCATED)
[0.9999477863311768,0.009791706688702106,-0.002925304928794503,0.0,-0.009806939400732517,0.999938189(...TRUNCATED)
[0.007755449041724205,-0.9999694228172302,-0.0010143029503524303,-0.007275538053363562,0.00229405588(...TRUNCATED)
[ 374, 1238 ]
injection
48
[ 22.723922729492188, 0.1656065434217453 ]
0.116695
257
256
000019
[67.61399841308594,8.795999526977539,2.515000104904175,0.0,68.66899871826172,9.152999877929688,2.552(...TRUNCATED)
19,624
4
[22.67790985107422,-1.2965974807739258,-1.0143458843231201,4.039999961853027,1.690000057220459,1.509(...TRUNCATED)
[ "Truck", "Car", "Van", "Car" ]
[5.750356674194336,3.1872334480285645,-0.45153993368148804,5.420000076293945,2.059999942779541,2.599(...TRUNCATED)
4
[ -1, 0, 0, -1 ]
[ 4796, 1219, 106, 19 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
49
[ 22.67790985107422, -1.2965974807739258 ]
0.07422
1,438
1,265
000020
[34.611000061035156,0.061000000685453415,1.378999948501587,0.1899999976158142,36.143001556396484,0.1(...TRUNCATED)
20,804
4
[27.00604820251465,0.5748535394668579,-0.5947405695915222,3.7100000381469727,1.5700000524520874,1.87(...TRUNCATED)
[ "Car" ]
[15.869402885437012,-2.6729772090911865,-0.8376937508583069,4.369999885559082,1.6100000143051147,1.3(...TRUNCATED)
1
[ 0 ]
[ 500 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
50
[ 27.00604820251465, 0.5748535394668579 ]
0.284069
290
266
000021
[65.99700164794922,3.6679999828338623,2.444000005722046,0.0,65.95899963378906,3.874000072479248,2.44(...TRUNCATED)
21,580
4
[17.52849006652832,-2.9411203861236572,-0.6883176565170288,4.070000171661377,1.690000057220459,1.679(...TRUNCATED)
[ "Cyclist", "Car", "Car", "Van", "Car", "Car", "Car", "Car" ]
[3.4309327602386475,-2.733675479888916,-0.9533327221870422,1.8899999856948853,0.5299999713897705,1.5(...TRUNCATED)
8
[ -1, 0, 1, 1, 0, 1, 2, 2 ]
[ 177, 833, 227, 972, 176, 112, 49, 26 ]
[721.5377197265625,0.0,609.559326171875,44.85728073120117,0.0,721.5377197265625,172.85400390625,0.21(...TRUNCATED)
[0.9999238848686218,0.009837759658694267,-0.007445048075169325,0.0,-0.00986979529261589,0.9999421238(...TRUNCATED)
[0.0075337449088692665,-0.9999713897705078,-0.00061660201754421,-0.004069766029715538,0.014802490361(...TRUNCATED)
[ 375, 1242 ]
injection
51
[ 17.52849006652832, -2.9411203861236572 ]
-0.166429
2,222
2,144
End of preview. Expand in Data Studio

ATLAS-KITTI

Adversarial LiDAR point clouds derived from the KITTI 3D object detection validation split. Anonymous release for peer review.

Configs

inject_easy, inject_medium, inject_hard, removal_az10, removal_az20, removal_az30, removal_az40, removal_az50, removal_az60

Every frame in a config is attacked; there are no clean frames.

Usage

from datasets import load_dataset
import numpy as np

ds = load_dataset("ps3020/atlas-kitti", "removal_az20", split="train")
ex = ds[0]

# points are stored FLAT -- reshape to recover the cloud
pts = np.asarray(ex["points"], np.float32).reshape(ex["num_points"], ex["point_dim"])
# (N, 4) = x, y, z, intensity

spoof_gt  = np.asarray(ex["spoof_gt"], np.float32)                      # (7,) attacked object
gt_boxes  = np.asarray(ex["gt_boxes"], np.float32).reshape(ex["num_gt"], 7)
gt_names  = ex["gt_names"]

Or use the bundled helper, which returns arrays directly:

from load_atlas_kitti import load_atlas, to_arrays, removal_rate

ds = load_atlas("removal_az20")                 # add streaming=True to avoid download
pts, spoof_gt, gt_boxes, gt_names = to_arrays(ds[0])

Attack success rate

spoof_gt is the attacked object. Score each frame by whether a prediction overlaps it at IoU >= 0.3:

  • injection — success = a detection appears (false positive created)
  • removal — success = the detection is missing (true positive destroyed)
asr = n_success / len(ds)

Use len(ds), not 3769. Frames that could not be attacked were never written, so configs differ in length:

config frames
inject_easy 3769
inject_medium 3767
inject_hard 3649
removal_az10removal_az60 3384

Fields

field description
frame_id KITTI frame id, e.g. "000001"
points, num_points, point_dim attacked cloud, flattened; reshape to (N, 4)
spoof_gt (7,) attacked object [x, y, z, dx, dy, dz, heading], LiDAR frame
gt_boxes, gt_names, num_gt clean KITTI ground truth (DontCare excluded)
gt_difficulty, gt_num_points KITTI difficulty, points per box
calib_P2, calib_R0_rect, calib_Tr_velo_to_cam calibration, flattened 4×4 row-major
image_shape [height, width] — varies across frames
atk_* attack parameters — 6 fields for injection, 20 for removal

For removal, the realised removal rate is atk_n_points_removed / atk_n_points_in_sector.

Calibration

ASR needs only spoof_gt (LiDAR frame). Calibration is included because the official KITTI 3D AP is computed in camera coordinates, so reproducing standard KITTI evaluation requires projecting predictions with these matrices:

P2  = np.asarray(ex["calib_P2"], np.float32).reshape(4, 4)
R0  = np.asarray(ex["calib_R0_rect"], np.float32).reshape(4, 4)
V2C = np.asarray(ex["calib_Tr_velo_to_cam"], np.float32).reshape(4, 4)
H, W = ex["image_shape"]

# LiDAR point/box centre -> image pixel
uv = P2 @ (R0 @ (V2C @ np.append(xyz, 1.0)))
uv = uv[:2] / uv[2]

Notes

  • Clouds are camera-FOV cropped (roughly |azimuth| <= 40°, x > 5 m), not full 360°.
  • Attacks target the Car class.
  • KITTI 3D object has no sequences, so frames are attacked independently.

License

cc-by-nc-sa-3.0, inherited from KITTI. Please cite KITTI alongside this dataset.

Downloads last month
-

Collection including ps3020/ATLAS-KITTI