Datasets:

tiu2mo
/

A2Seek

anomaly_name stringclasses 19 values	scene stringclasses 10 values	lighting stringclasses 7 values	period stringclasses 3 values	altitude stringclasses 2 values	velocity stringclasses 6 values	sample_id stringlengths 29 29	source_split stringclasses 3 values	scene_code stringclasses 10 values	code stringlengths 27 27	objects_json stringlengths 37 122
Normal	sidewalk	sunny	afternoon	10-20(m)	0-10(m/s)	video_group_021/video_1/06336	validation	S02	E00S02L0P0A0V3_G021I1F06336	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Animal	school_gate	rainy	afternoon	20-50(m)	0-3(m/s)	video_group_067/video_2/04352	train	S07	E04S07L3P0A1V0_G067I2F04352	{"ref":["Animal"],"bbox":[[569,403,597,447]],"region":[504,308,661,448]}
Unconventional Vehicle	open_space	overcast	morning	10-20(m)	0-3(m/s)	video_group_230/video_4/06272	train	S04	E01S04L2P1A0V0_G230I4F06272	{"ref":["Unconventional Vehicle","Treading on Grass"],"bbox":[[371,31,387,57],[64,44,119,88]],"region":[301,0,457,139]}
Loitering	wall	dark_light	night	10-20(m)	3-10(m/s)	video_group_045/video_1/01984	validation	S08	E01S08L8P2A0V1_G045I1F01984	{"ref":["Loitering"],"bbox":[[110,10,163,54]],"region":[58,0,214,139]}
Running	other	overcast	afternoon	20-50(m)	0-3(m/s)	video_group_249/video_2/02048	test	S00	E03S00L2P0A1V0_G249I2F02048	{"ref":["Running","Fighting"],"bbox":[[281,166,295,177],[281,166,295,177]],"region":[244,88,401,228]}
Falling	playground	dark_light	night	10-20(m)	0-3(m/s)	video_group_216/video_1/04864	validation	S03	E06S03L8P2A0V0_G216I1F04864	{"ref":["Falling"],"bbox":[[67,313,112,338]],"region":[11,256,168,395]}
Normal	sidewalk	cloudy	morning	20-50(m)	0-20(m/s)	video_group_096/video_1/08128	train	S02	E12S02L1P1A1V5_G096I1F08128	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Animal	school_gate	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_025/video_1/08448	validation	S07	E04S07L0P0A1V0_G025I1F08448	{"ref":["Animal"],"bbox":[[406,312,427,332]],"region":[338,252,494,392]}
Robbery	open_space	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_181/video_9/01280	train	S04	E19S04L0P0A0V0_G181I9F01280	{"ref":["Robbery"],"bbox":[[378,228,419,256]],"region":[312,177,469,316]}
Loitering	wall	dark_light	night	20-50(m)	3-10(m/s)	video_group_048/video_2/02048	train	S08	E01S08L8P2A1V1_G048I2F02048	{"ref":["Loitering"],"bbox":[[58,154,221,356]],"region":[26,154,253,356]}
Running	road	rainy	morning	20-50(m)	3-10(m/s)	video_group_159/video_3/01152	validation	S01	E03S01L3P1A1V1_G159I3F01152	{"ref":["Running"],"bbox":[[395,98,409,122]],"region":[336,24,493,164]}
Lost Item	park	overcast	afternoon	20-50(m)	3-10(m/s)	video_group_054/video_4/00832	test	S05	E02S05L2P0A1V1_G054I4F00832	{"ref":["Lost Item"],"bbox":[[238,11,262,29]],"region":[171,0,328,139]}
Normal	teaching_building	foggy	morning	20-50(m)	0-3(m/s)	video_group_077/video_1/05504	test	S09	E00S09L5P1A1V0_G077I1F05504	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Bicycling	sidewalk	dark_light	night	10-20(m)	0-3(m/s)	video_group_114/video_1/00896	train	S02	E13S02L8P2A0V0_G114I1F00896	{"ref":["Bicycling"],"bbox":[[551,131,602,162]],"region":[498,77,655,216]}
Robbery	playground	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_164/video_1/06400	validation	S03	E19S03L0P0A0V0_G164I1F06400	{"ref":["Robbery"],"bbox":[[400,39,429,115]],"region":[336,7,493,146]}
Loitering	wall	twilight	morning	20-50(m)	0-3(m/s)	video_group_041/video_1/03392	validation	S08	E01S08L7P1A1V0_G041I1F03392	{"ref":["Loitering"],"bbox":[[0,177,135,442]],"region":[0,177,298,442]}
Unconventional Vehicle	road	rainy	morning	10-20(m)	3-10(m/s)	video_group_158/video_3/03072	train	S01	E09S01L3P1A0V1_G158I3F03072	{"ref":["Unconventional Vehicle"],"bbox":[[633,377,675,439]],"region":[576,308,732,448]}
Treading on Grass	park	rainy	morning	10-20(m)	0-3(m/s)	video_group_228/video_2/03136	test	S05	E02S05L3P1A0V0_G228I2F03136	{"ref":["Treading on Grass"],"bbox":[[402,225,489,281]],"region":[367,184,524,323]}
Normal	rooftop	sunny	morning	20-50(m)	3-10(m/s)	video_group_027/video_1/02112	test	S06	E00S06L0P1A1V1_G027I1F02112	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Running	sidewalk	overcast	afternoon	10-20(m)	3-10(m/s)	video_group_245/video_1/01664	train	S02	E03S02L2P0A0V1_G245I1F01664	{"ref":["Running"],"bbox":[[575,284,592,313]],"region":[505,228,662,368]}
Fighting	other	overcast	afternoon	20-50(m)	0-3(m/s)	video_group_251/video_1/02176	test	S00	E18S00L2P0A1V0_G251I1F02176	{"ref":["Fighting"],"bbox":[[394,308,419,327]],"region":[329,247,486,387]}
Robbery	playground	sunny	afternoon	10-20(m)	3-10(m/s)	video_group_175/video_1/01984	validation	S03	E19S03L0P0A0V1_G175I1F01984	{"ref":["Robbery"],"bbox":[[431,183,458,208]],"region":[366,143,522,282]}
Playing with Water	road	rainy	morning	10-20(m)	3-10(m/s)	video_group_222/video_1/03840	train	S01	E12S01L3P1A0V1_G222I1F03840	{"ref":["Playing with Water"],"bbox":[[779,388,814,433]],"region":[718,308,875,448]}
Treading on Grass	park	rainy	morning	10-20(m)	0-3(m/s)	video_group_228/video_1/04160	test	S05	E02S05L3P1A0V0_G228I1F04160	{"ref":["Treading on Grass"],"bbox":[[545,142,581,191]],"region":[484,97,641,236]}
Normal	teaching_building	foggy	afternoon	20-50(m)	0-3(m/s)	video_group_080/video_1/03456	test	S09	E00S09L5P0A1V0_G080I1F03456	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Running	school_gate	overcast	afternoon	10-20(m)	0-3(m/s)	video_group_255/video_1/04800	validation	S07	E04S07L2P0A0V0_G255I1F04800	{"ref":["Running"],"bbox":[[413,225,425,253]],"region":[289,186,446,326]}
Jaywalking	sidewalk	dark_light	night	10-20(m)	0-3(m/s)	video_group_117/video_1/08640	validation	S02	E12S02L8P2A0V0_G117I1F08640	{"ref":["Jaywalking"],"bbox":[[399,212,413,232]],"region":[328,152,485,291]}
Carrying Weapon	open_space	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_263/video_1/01664	validation	S04	E11S04L0P0A1V0_G263I1F01664	{"ref":["Carrying Weapon","Fighting"],"bbox":[[470,179,484,275],[468,162,479,182]],"region":[399,157,555,296]}
Wall Climbing	playground	cloudy	afternoon	10-20(m)	0-3(m/s)	video_group_261/video_1/03200	train	S03	E10S03L1P0A0V0_G261I1F03200	{"ref":["Wall Climbing"],"bbox":[[511,140,553,210]],"region":[453,105,610,245]}
Playing with Water	park	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_186/video_1/07040	test	S05	E08S05L0P0A1V0_G186I1F07040	{"ref":["Playing with Water"],"bbox":[[438,227,473,263]],"region":[377,175,534,314]}
Normal	rooftop	sunny	morning	20-50(m)	3-10(m/s)	video_group_027/video_3/00768	test	S06	E00S06L0P1A1V1_G027I3F00768	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Loitering	wall	twilight	morning	20-50(m)	10-20(m/s)	video_group_211/video_1/01344	train	S08	E01S08L7P1A1V2_G211I1F01344	{"ref":["Loitering"],"bbox":[[92,182,207,296]],"region":[71,169,228,308]}
Jaywalking	road	rainy	morning	10-20(m)	3-10(m/s)	video_group_157/video_2/00384	train	S01	E12S01L3P1A0V1_G157I2F00384	{"ref":["Jaywalking"],"bbox":[[857,196,895,240]],"region":[739,148,896,287]}
Carrying Weapon	other	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_269/video_1/01088	test	S00	E20S00L0P0A0V0_G269I1F01088	{"ref":["Carrying Weapon","Bullying"],"bbox":[[372,202,421,248],[354,194,434,248]],"region":[294,138,450,278]}
Unable to Stand	playground	cloudy	afternoon	20-50(m)	0-3(m/s)	video_group_123/video_1/00832	validation	S03	E07S03L1P0A1V0_G123I1F00832	{"ref":["Unable to Stand"],"bbox":[[121,387,137,404]],"region":[51,308,207,448]}
Bicycling	sidewalk	dark_light	night	20-50(m)	3-10(m/s)	video_group_112/video_1/02688	train	S02	E13S02L8P2A1V1_G112I1F02688	{"ref":["Bicycling"],"bbox":[[130,321,171,361]],"region":[72,271,229,411]}
Normal	rooftop	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_024/video_2/01152	test	S06	E00S06L0P0A1V0_G024I2F01152	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Robbery	open_space	sunny	morning	20-50(m)	0-3(m/s)	video_group_161/video_1/01984	validation	S04	E19S04L0P1A1V0_G161I1F01984	{"ref":["Robbery"],"bbox":[[427,219,453,249]],"region":[361,164,518,303]}
Jaywalking	road	rainy	morning	10-20(m)	3-10(m/s)	video_group_157/video_1/05312	train	S01	E12S01L3P1A0V1_G157I1F05312	{"ref":["Jaywalking"],"bbox":[[536,349,561,392]],"region":[470,301,627,440]}
Animal	park	rainy	night	20-50(m)	3-10(m/s)	video_group_210/video_1/01728	test	S05	E04S05L3P2A1V1_G210I1F01728	{"ref":["Animal"],"bbox":[[468,422,480,435]],"region":[396,308,553,448]}
Theft	wall	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_163/video_1/03584	train	S08	E17S08L0P0A1V0_G163I1F03584	{"ref":["Theft"],"bbox":[[198,327,219,371]],"region":[130,280,287,419]}
Fighting	sidewalk	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_191/video_1/00448	validation	S02	E18S02L0P0A1V0_G191I1F00448	{"ref":["Fighting"],"bbox":[[546,224,589,271]],"region":[489,178,646,317]}
Normal	other	dark_light	night	20-50(m)	0-3(m/s)	video_group_011/video_1/03712	test	S00	E00S00L8P2A1V0_G011I1F03712	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Sneaking	open_space	sunny	morning	10-20(m)	0-3(m/s)	video_group_233/video_2/04224	train	S04	E16S04L0P1A0V0_G233I2F04224	{"ref":["Sneaking"],"bbox":[[515,217,549,261]],"region":[454,169,611,308]}
Jaywalking	road	dark_light	night	20-50(m)	0-3(m/s)	video_group_129/video_1/02048	validation	S01	E12S01L8P2A1V0_G129I1F02048	{"ref":["Jaywalking"],"bbox":[[474,182,521,238]],"region":[419,140,576,280]}
Lost Item	park	overcast	afternoon	20-50(m)	3-10(m/s)	video_group_054/video_1/00832	test	S05	E09S05L2P0A1V1_G054I1F00832	{"ref":["Lost Item"],"bbox":[[241,0,288,26]],"region":[186,0,343,139]}
Loitering	wall	twilight	morning	10-20(m)	3-10(m/s)	video_group_039/video_1/02048	train	S08	E01S08L7P1A0V1_G039I1F02048	{"ref":["Loitering"],"bbox":[[0,90,111,181]],"region":[0,65,156,205]}
Running	sidewalk	dark_light	night	10-20(m)	0-3(m/s)	video_group_215/video_1/00320	validation	S02	E03S02L8P2A0V0_G215I1F00320	{"ref":["Running"],"bbox":[[459,219,488,256]],"region":[395,168,552,308]}
Normal	teaching_building	foggy	morning	10-20(m)	0-3(m/s)	video_group_076/video_1/05056	test	S09	E00S09L5P1A0V0_G076I1F05056	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Carrying Weapon	open_space	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_184/video_3/01216	train	S04	E11S04L0P0A0V0_G184I3F01216	{"ref":["Carrying Weapon"],"bbox":[[431,159,471,184]],"region":[382,102,539,241]}
Jaywalking	road	dark_light	night	10-20(m)	0-3(m/s)	video_group_132/video_2/06720	train	S01	E12S01L8P2A0V0_G132I2F06720	{"ref":["Jaywalking"],"bbox":[[280,90,293,115]],"region":[208,32,365,172]}
Playing with Water	park	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_186/video_2/03328	test	S05	E08S05L0P0A1V0_G186I2F03328	{"ref":["Playing with Water"],"bbox":[[417,143,512,187]],"region":[417,53,707,311]}
Loitering	wall	twilight	morning	10-20(m)	0-3(m/s)	video_group_038/video_2/03136	validation	S08	E01S08L7P1A0V0_G038I2F03136	{"ref":["Loitering"],"bbox":[[181,0,232,41]],"region":[128,0,284,139]}
Lost Item	school_gate	cloudy	afternoon	10-20(m)	3-10(m/s)	video_group_259/video_2/01024	train	S07	E01S07L1P0A0V1_G259I2F01024	{"ref":["Lost Item"],"bbox":[[571,259,600,286]],"region":[507,202,664,342]}
Normal	rooftop	rainy	afternoon	20-50(m)	0-3(m/s)	video_group_066/video_1/04160	test	S06	E00S06L3P0A1V0_G066I1F04160	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Fighting	open_space	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_184/video_1/03072	validation	S04	E18S04L0P0A0V0_G184I1F03072	{"ref":["Fighting","Loitering"],"bbox":[[492,169,515,219],[763,338,824,380]],"region":[427,116,583,256]}
Running	road	overcast	morning	10-20(m)	3-10(m/s)	video_group_229/video_1/02688	train	S01	E03S01L2P1A0V1_G229I1F02688	{"ref":["Running"],"bbox":[[457,72,466,96]],"region":[383,14,540,154]}
Playing with Water	park	sunny	morning	10-20(m)	0-3(m/s)	video_group_231/video_2/01984	test	S05	E08S05L0P1A0V0_G231I2F01984	{"ref":["Playing with Water"],"bbox":[[484,142,501,230]],"region":[414,116,571,255]}
Treading on Grass	wall	twilight	morning	20-50(m)	3-10(m/s)	video_group_214/video_1/01152	train	S08	E01S08L7P1A1V1_G214I1F01152	{"ref":["Treading on Grass"],"bbox":[[830,301,853,319]],"region":[739,240,896,379]}
Unable to Stand	playground	cloudy	afternoon	10-20(m)	0-3(m/s)	video_group_262/video_1/02624	train	S03	E07S03L1P0A0V0_G262I1F02624	{"ref":["Unable to Stand"],"bbox":[[445,343,501,378]],"region":[394,291,551,430]}
Normal	rooftop	rainy	afternoon	20-50(m)	0-3(m/s)	video_group_066/video_2/05696	test	S06	E00S06L3P0A1V0_G066I2F05696	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Lost Item	open_space	cloudy	afternoon	10-20(m)	0-3(m/s)	video_group_258/video_1/01664	validation	S04	E01S04L1P0A0V0_G258I1F01664	{"ref":["Lost Item"],"bbox":[[249,385,261,395]],"region":[177,308,333,448]}
Robbery	road	overcast	afternoon	10-20(m)	0-3(m/s)	video_group_246/video_1/00960	train	S01	E17S01L2P0A0V0_G246I1F00960	{"ref":["Robbery"],"bbox":[[654,221,683,264]],"region":[590,169,746,308]}
Playing with Water	park	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_186/video_4/04544	test	S05	E08S05L0P0A1V0_G186I4F04544	{"ref":["Playing with Water"],"bbox":[[428,93,470,162]],"region":[371,57,527,197]}
Loitering	wall	twilight	morning	20-50(m)	10-20(m/s)	video_group_043/video_2/00064	train	S08	E01S08L7P1A1V2_G043I2F00064	{"ref":["Loitering"],"bbox":[[342,36,380,72]],"region":[269,36,527,265]}
Unable to Stand	playground	cloudy	afternoon	20-50(m)	10-20(m/s)	video_group_125/video_2/01024	train	S03	E07S03L1P0A1V2_G125I2F01024	{"ref":["Unable to Stand"],"bbox":[[830,314,864,336]],"region":[739,255,896,395]}
Normal	teaching_building	foggy	afternoon	10-20(m)	0-3(m/s)	video_group_081/video_1/08256	test	S09	E00S09L5P0A0V0_G081I1F08256	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Lost Item	open_space	cloudy	afternoon	10-20(m)	0-3(m/s)	video_group_258/video_2/02496	train	S04	E01S04L1P0A0V0_G258I2F02496	{"ref":["Lost Item","Unconventional Vehicle"],"bbox":[[429,277,439,289],[135,208,162,241]],"region":[356,214,513,353]}
Fighting	road	overcast	afternoon	10-20(m)	0-3(m/s)	video_group_248/video_1/02368	train	S01	E03S01L2P0A0V0_G248I1F02368	{"ref":["Fighting"],"bbox":[[546,204,571,229]],"region":[478,146,635,286]}
Playing with Water	park	sunny	morning	10-20(m)	0-3(m/s)	video_group_231/video_1/03648	test	S05	E08S05L0P1A0V0_G231I1F03648	{"ref":["Playing with Water"],"bbox":[[420,200,480,247]],"region":[371,154,528,293]}
Treading on Grass	wall	twilight	morning	10-20(m)	10-20(m/s)	video_group_040/video_2/00640	validation	S08	E01S08L7P1A0V2_G040I2F00640	{"ref":["Treading on Grass"],"bbox":[[753,159,775,192]],"region":[686,106,842,245]}
Unable to Stand	playground	cloudy	afternoon	20-50(m)	10-20(m/s)	video_group_125/video_3/00128	validation	S03	E07S03L1P0A1V2_G125I3F00128	{"ref":["Unable to Stand"],"bbox":[[511,159,527,172]],"region":[440,96,597,235]}
Normal	other	dark_light	night	20-50(m)	0-20(m/s)	video_group_089/video_2/00000	test	S00	E00S00L8P2A1V5_G089I2F00000	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Running	open_space	overcast	morning	10-20(m)	0-3(m/s)	video_group_230/video_2/00960	train	S04	E03S04L2P1A0V0_G230I2F00960	{"ref":["Running"],"bbox":[[830,84,872,123]],"region":[739,33,896,172]}
Carrying Weapon	road	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_268/video_1/01088	train	S01	E11S01L0P0A0V0_G268I1F01088	{"ref":["Carrying Weapon","Running"],"bbox":[[493,170,525,209],[375,197,422,237]],"region":[430,120,587,259]}
Lost Item	park	overcast	afternoon	20-50(m)	3-10(m/s)	video_group_054/video_2/01536	test	S05	E02S05L2P0A1V1_G054I2F01536	{"ref":["Lost Item"],"bbox":[[361,187,419,288]],"region":[311,168,468,308]}
Falling	sidewalk	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_192/video_1/00960	validation	S02	E03S02L0P0A1V0_G192I1F00960	{"ref":["Falling"],"bbox":[[459,213,485,232]],"region":[394,153,550,292]}
Fighting	school_gate	overcast	afternoon	10-20(m)	0-3(m/s)	video_group_257/video_1/02176	train	S07	E03S07L2P0A0V0_G257I1F02176	{"ref":["Fighting"],"bbox":[[485,253,504,280]],"region":[425,202,582,342]}
Normal	rooftop	overcast	morning	20-50(m)	0-3(m/s)	video_group_010/video_2/02752	test	S06	E00S06L2P1A1V0_G010I2F02752	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Theft	open_space	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_183/video_3/04736	train	S04	E17S04L0P0A1V0_G183I3F04736	{"ref":["Theft"],"bbox":[[423,314,442,336]],"region":[354,255,511,395]}
Running	road	sunny	morning	20-50(m)	0-3(m/s)	video_group_156/video_2/10368	train	S01	E12S01L0P1A1V0_G156I2F10368	{"ref":["Running"],"bbox":[[658,172,675,198]],"region":[588,115,745,255]}
Playing with Water	park	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_186/video_3/06464	test	S05	E02S05L0P0A1V0_G186I3F06464	{"ref":["Playing with Water"],"bbox":[[392,198,428,230]],"region":[350,131,507,271]}
Jaywalking	sidewalk	dark_light	night	10-20(m)	10-20(m/s)	video_group_119/video_1/00320	train	S02	E12S02L8P2A0V2_G119I1F00320	{"ref":["Jaywalking"],"bbox":[[252,104,330,142]],"region":[212,53,369,192]}
Lost Item	school_gate	cloudy	afternoon	10-20(m)	3-10(m/s)	video_group_259/video_1/01728	train	S07	E01S07L1P0A0V1_G259I1F01728	{"ref":["Lost Item"],"bbox":[[498,217,511,237]],"region":[426,157,583,296]}
Normal	rooftop	overcast	morning	20-50(m)	0-3(m/s)	video_group_010/video_1/04032	test	S06	E00S06L2P1A1V0_G010I1F04032	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Theft	open_space	sunny	afternoon	20-50(m)	0-3(m/s)	video_group_183/video_8/03648	validation	S04	E12S04L0P0A1V0_G183I8F03648	{"ref":["Theft"],"bbox":[[439,158,472,182]],"region":[377,100,534,240]}
Treading on Grass	wall	sunny	afternoon	20-50(m)	3-10(m/s)	video_group_201/video_1/00256	train	S08	E02S08L0P0A1V1_G201I1F00256	{"ref":["Treading on Grass"],"bbox":[[238,28,277,65]],"region":[179,0,336,139]}
Playing with Water	park	overcast	afternoon	10-20(m)	0-3(m/s)	video_group_207/video_1/06656	test	S05	E08S05L2P0A0V0_G207I1F06656	{"ref":["Playing with Water","Playing with Water"],"bbox":[[135,287,158,317],[122,278,175,321]],"region":[68,232,225,372]}
Jaywalking	road	dark_light	night	10-20(m)	0-3(m/s)	video_group_132/video_1/00512	validation	S01	E12S01L8P2A0V0_G132I1F00512	{"ref":["Jaywalking"],"bbox":[[462,202,476,233]],"region":[390,148,547,288]}
Bicycling	school_gate	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_022/video_2/03520	validation	S07	E01S07L0P0A0V0_G022I2F03520	{"ref":["Bicycling"],"bbox":[[256,152,281,179]],"region":[190,96,347,235]}
Normal	rooftop	dark_light	night	20-50(m)	0-3(m/s)	video_group_075/video_2/06208	test	S06	E09S06L8P2A1V0_G075I2F06208	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Falling	open_space	cloudy	afternoon	20-50(m)	0-3(m/s)	video_group_239/video_1/01792	train	S04	E06S04L1P0A1V0_G239I1F01792	{"ref":["Falling"],"bbox":[[704,34,731,54]],"region":[639,0,795,139]}
Animal	wall	overcast	afternoon	20-50(m)	0-3(m/s)	video_group_208/video_1/00384	train	S08	E01S08L2P0A1V0_G208I1F00384	{"ref":["Animal"],"bbox":[[249,22,263,32]],"region":[178,0,335,139]}
Playing with Water	sidewalk	overcast	afternoon	10-20(m)	0-3(m/s)	video_group_202/video_1/03776	train	S02	E08S02L2P0A0V0_G202I1F03776	{"ref":["Playing with Water"],"bbox":[[463,0,510,88]],"region":[408,0,564,139]}
Lost Item	park	overcast	afternoon	10-20(m)	10-20(m/s)	video_group_052/video_5/01472	test	S05	E09S05L2P0A0V2_G052I5F01472	{"ref":["Lost Item"],"bbox":[[287,242,305,263]],"region":[217,182,374,322]}
Bicycling	school_gate	cloudy	afternoon	10-20(m)	0-3(m/s)	video_group_013/video_1/02240	validation	S07	E13S07L1P0A0V0_G013I1F02240	{"ref":["Bicycling"],"bbox":[[350,92,374,115]],"region":[283,34,440,173]}
Normal	rooftop	rainy	afternoon	20-50(m)	3-10(m/s)	video_group_065/video_2/01088	test	S06	E00S06L3P0A1V1_G065I2F01088	{"ref":["Normal"],"bbox":[[0,0,0,0]]}
Sneaking	open_space	sunny	morning	10-20(m)	0-3(m/s)	video_group_233/video_1/00192	validation	S04	E16S04L0P1A0V0_G233I1F00192	{"ref":["Sneaking"],"bbox":[[396,187,427,214]],"region":[333,131,490,270]}
Loitering	teaching_building	foggy	afternoon	20-50(m)	0-3(m/s)	video_group_078/video_1/09472	test	S09	E01S09L5P0A1V0_G078I1F09472	{"ref":["Loitering"],"bbox":[[14,309,100,360]],"region":[0,265,156,404]}
Robbery	playground	sunny	afternoon	10-20(m)	0-3(m/s)	video_group_179/video_1/04160	train	S03	E19S03L0P0A0V0_G179I1F04160	{"ref":["Robbery"],"bbox":[[492,172,528,213]],"region":[431,123,588,262]}

End of preview. Expand in Data Studio

A2Seek

Dataset for aerial anomaly understanding, spatial grounding and structured visual reasoning.

Paper · Project

Overview of A2Seek (Figure 1 from the paper).

Dataset contents

The dataset has 8,192 sample IDs: 3,489 training, 2,327 validation and 2,376 test examples. The cot_single_reflection and raw configurations provide training targets and base annotations for the same sample IDs.

The base annotations reference 32,768 distinct frames. Training/validation cover scene codes 01/02/03/04/07/08; test covers 00/05/06/09. Splits are disjoint by sample, with 383 videos shared between train and validation.

data/ contains Hugging Face sample tables. dataset/ contains native JSONL files for training. annotations/ contains temporal intervals, tracks, human descriptions, initial/revised frame annotations and teacher reasoning targets.

metadata/splits/ contains the saved train/validation/test sample ID lists and ID-train/ID-eval/OOD-eval aliases. metadata/domain_protocol.json records domain scene codes, split counts and annotation/list checksums. Sample membership and order are preserved.

Images and video indices

175,146 JPEG images are stored losslessly in 115 Parquet shards under data/images/. The preview configuration displays 200 representative examples spanning scenes, behavior categories and capture conditions. It is a demonstration gallery; official sample membership is defined by the annotation configurations and split lists. The full image bank remains available under data/images/.

Full-frame training views use 896 × 448 images; crop views use 448 × 448. Image paths are relative to the dataset repository root. Bounding boxes use xyxy pixel coordinates in the corresponding image view.

metadata/canonical_video_index.jsonl describes annotated clips. metadata/raw_capture_index.jsonl records original capture filenames, camera suffixes and stream parameters. Capture candidates are grouped by video group; the canonical_clip_match/raw_capture_mapping fields record the match status. Source roots are configured locally. Original videos stay in their source location.

metadata/dji_clip_mapping.jsonl provides 542 one-to-one correspondences between encoded annotated clips and original W-camera filenames/relative paths. Each is verified using frame count and image content. metadata/dji_video_group_mapping.jsonl indexes all 1,616 captures: 542 verified_clip records, 1,072 paired_capture records for Z/T/S streams sharing the capture directory and DJI timestamp/sequence, and two additional group_candidate records. Companion streams retain their own views and frame timing.

metadata/video_annotation_index.jsonl contains 542 encoded clip records linked to original DJI filenames. For example, video_group_001/video_1 corresponds to video_group_001_@_video_1.json, frame directory video_group_001/video_1, training-image directory images/video_group_001_@_video_1, and DJI_20231216164143_0001_W.MP4. The index includes the annotation attribute code and relative paths to the initial/revised snapshots. The (video_group, video_name) pair uniquely identifies a clip; event/environment attribute codes can repeat across clips.

Schema

Each annotation row contains sample_id, group/video/frame IDs, code, anomaly metadata, images, messages, objects_json and nullable solution. Use json.loads(row["objects_json"]) to recover object labels and boxes. messages contains reference training targets; inference uses the user messages. Media paths are strings and can be opened relative to a downloaded snapshot.

Loading

from datasets import load_dataset
dataset = load_dataset("tiu2mo/A2Seek", "cot_single_reflection")

For a local checkout, pass the three files under data/CONFIG/ to load_dataset("parquet", data_files=...) or the JSON loader for a JSONL export. See annotation_documentation.md for the native layouts and coordinates.

Configurations

Configuration	Contents	Splits
preview	200 representative examples with image previews	preview
cot_single_reflection	Complete training targets	train: 3489; validation: 2327; test: 2376
raw	Complete base annotations	train: 3489; validation: 2327; test: 2376

Provenance

source_manifest.json records original annotation filenames, transformations and SHA-256 hashes. export_manifest.json records the complete upload inventory. metadata/image_manifest.jsonl records image checksums and dimensions when included. Supporting teacher-supervision records may overlap; configuration rows are alternative views of the same samples.

Citation

Use the included CITATION.bib or CITATION.cff to cite A2Seek (NeurIPS 2025).

Native training setup

Download the dataset snapshot, then restore its image paths from the A2Seek code checkout:

export A2SEEK_DATA_ROOT=/path/to/A2Seek-dataset
uv run --no-sync hf download tiu2mo/A2Seek --repo-type dataset --local-dir "$A2SEEK_DATA_ROOT"
uv run --no-sync python scripts/data/prepare_local_data.py

Preparation restores the original JPEG bytes under images/ before resolving native JSONL paths. scripts/data/materialize_images.py also performs this step independently.

To stream all image records:

images = load_dataset("parquet",
    data_files="hf://datasets/tiu2mo/A2Seek/data/images/train-*.parquet",
    split="train", streaming=True)
first = next(iter(images))
image = first["image"]

Downloads last month: 33

Paper for tiu2mo/A2Seek

A2Seek: Towards Reasoning-Centric Benchmark for Aerial Anomaly Understanding

Paper • 2505.21962 • Published Nov 25, 2025