Support output to OpenPose JSON format #650
Assignees
Comments
|
Two other standards are:
I can send you samples of OpenPose, deeplabcut, and Coco files if you are interested. |
|
Thank you. I am interested so if those are easily accessible it would be awesome to have some references. Mac |
|
Here are some sample data (not from the same trials): OpenPose format (one file per frame, keypoint order depends on model, coordinates x,y,conf) {
"people": [
{
"face_keypoints_2d": [
],
"face_keypoints_3d": [
],
"hand_left_keypoints_2d": [
],
"hand_left_keypoints_3d": [
],
"hand_right_keypoints_2d": [
],
"hand_right_keypoints_3d": [
],
"person_id": [
-1
],
"pose_keypoints_2d": [
785.032272338867,
874.429085731506,
0.930248200893402,
796.808270772298,
862.653086662292,
0.927412450313568,
799.752270380656,
862.653086662292,
0.861262261867523,
814.472268422445,
895.037084102631,
0.902777314186096,
902.792256673177,
903.869083404541,
0.949171543121338,
743.816277821859,
1051.06907176971,
0.765520691871643,
935.176252365112,
1036.3490729332,
0.809015989303589,
702.60028330485,
1227.70905780792,
0.772481501102448,
905.736256281535,
1207.1010594368,
0.780025482177734,
605.448296229044,
1333.69304943085,
0.793694794178009,
835.080265680949,
1260.09305524826,
0.558070302009583,
699.656283696493,
1339.58104896545,
0.680959105491638,
808.584269205729,
1342.52504873276,
0.684390842914581,
640.776291529337,
1525.05303430557,
0.8237544298172,
746.760277430217,
1486.78103733063,
0.782606363296509,
581.896299362183,
1692.86102104187,
0.868499636650085,
705.544282913208,
1648.70102453232,
0.813148558139801,
855.688262939453,
794.941092014313,
0.893165171146393,
849.800263722738,
962.74907875061,
0.913672804832458,
761.480275472005,
1321.91705036163,
1.05537974834442,
508.296309153239,
1636.9250254631,
0.869200706481934,
652.552289962769,
1616.31702709198,
0.742750108242035,
502.408309936523,
1666.36502313614,
0.921308159828186,
687.880285263062,
1628.09302616119,
0.713791012763977,
584.840298970541,
1728.18901824951,
0.935503244400024,
705.544282913208,
1675.19702243805,
0.864744603633881
],
"pose_keypoints_3d": [
]
},
{
"face_keypoints_2d": [
],
"face_keypoints_3d": [
],
"hand_left_keypoints_2d": [
],
"hand_left_keypoints_3d": [
],
"hand_right_keypoints_2d": [
],
"hand_right_keypoints_3d": [
],
"person_id": [
-1
],
"pose_keypoints_2d": [
493.176073074341,
434.241302490234,
0.719350218772888,
503.458368301392,
421.388433456421,
0.783019661903381,
480.323204040527,
421.388433456421,
0.755906939506531,
511.17008972168,
423.959007263184,
0.819137573242188,
444.33517074585,
413.676712036133,
0.867408633232117,
521.45238494873,
501.076221466064,
0.770452260971069,
369.788530349731,
485.652778625488,
0.772427380084991,
539.446401596069,
603.899173736572,
0.775422871112823,
287.530168533325,
596.187452316284,
0.830751299858093,
549.72869682312,
709.292699813843,
0.765416979789734,
261.824430465698,
719.574995040894,
0.820626974105835,
470.040908813477,
732.427864074707,
0.74791944026947,
377.50025177002,
727.286716461182,
0.713379383087158,
457.188039779663,
912.368030548096,
0.840719163417816,
364.647382736206,
902.085735321045,
0.842771649360657,
421.200006484985,
1087.16704940796,
0.855790734291077,
333.800497055054,
1007.47926139832,
0.843364059925079,
500.887794494629,
339.130071640015,
0.81070351600647,
457.188039779663,
462.517614364624,
0.845634043216705,
423.770580291748,
724.716142654419,
1.06806135177612,
436.623449325562,
1151.43139457703,
0.824704349040985,
341.512218475342,
1069.17303276062,
0.773277282714844,
452.046892166138,
1143.71967315674,
0.841744661331177,
318.377054214478,
1058.89073753357,
0.780112028121948,
403.205989837646,
1105.1610660553,
0.787861347198486,
328.659349441528,
1010.04983520508,
0.746102213859558
],
"pose_keypoints_3d": [
]
},
{
"face_keypoints_2d": [
],
"face_keypoints_3d": [
],
"hand_left_keypoints_2d": [
],
"hand_left_keypoints_3d": [
],
"hand_right_keypoints_2d": [
],
"hand_right_keypoints_3d": [
],
"person_id": [
-1
],
"pose_keypoints_2d": [
79.9320793151855,
343.463024139404,
0.760430693626404,
87.3958969116211,
334.506443023682,
0.80179488658905,
72.46826171875,
333.013679504395,
0.811470806598663,
99.338005065918,
338.984733581543,
0.828160524368286,
53.0623359680176,
334.506443023682,
0.82732355594635,
117.251167297363,
388.245929718018,
0.792685210704803,
30.6708831787109,
382.274875640869,
0.776597559452057,
111.280113220215,
456.913051605225,
0.784640848636627,
11.2649574279785,
449.449234008789,
0.736559152603149,
72.46826171875,
404.666328430176,
0.806424021720886,
6.78666687011719,
506.174247741699,
0.650886654853821,
81.4248428344727,
518.116355895996,
0.690829515457153,
35.1491737365723,
515.130828857422,
0.713027656078339,
76.9465522766113,
625.595329284668,
0.824519991874695,
30.6708831787109,
619.62427520752,
0.814788639545441,
72.46826171875,
721.132194519043,
0.812629520893097,
29.1781196594238,
707.697322845459,
0.802818536758423,
76.9465522766113,
291.216300964355,
0.772570133209229,
73.9610252380371,
365.854476928711,
0.882087588310242,
59.033390045166,
515.130828857422,
0.980525076389313,
50.0768089294434,
750.987464904785,
0.787731468677521,
14.2504844665527,
739.045356750488,
0.781021475791931,
66.4972076416016,
753.972991943359,
0.787867844104767,
3.80113983154297,
734.567066192627,
0.780524611473083,
72.46826171875,
736.059829711914,
0.763829290866852,
35.1491737365723,
725.610485076904,
0.779185891151428
],
"pose_keypoints_3d": [
]
}
],
"version": 1.3
}DeepLabCut single animal format (one file for all frames)DeepLabCut multi animal format (one file for all frames)MMPose format (one file for all frames). I don't have any file handy, but here is the structure (see there) |
|
This is incredibly helpful to have as a reference, thank you! It's great to see how others approach the problem of storing this data. An ecosystem of open source tools that can interoperate to facilitate mocap is near and dear to my heart, so this is much appreciated. Mac |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
OpenPose provides a standard JSON format for its output:
https://cmu-perceptual-computing-lab.github.io/openpose/web/html/doc/md_doc_02_output.html
Rather than attempting to create some new style of representation, it may be best to default to a previously existing format that has survived some tire-kicking.
Note that OpenPose links to a format for representing the camera matrix, but the link is currently 404'd: https://cmu-perceptual-computing-lab.github.io/openpose/web/html/doc/advanced/calibration_module.md
The text was updated successfully, but these errors were encountered: