dexmulti/cameras.json

{
 "convention": "fx_fy_cx_cy: pinhole intrinsics of the 640x480 color image; T_base_cam: 4x4 transform mapping points from the camera optical frame (OpenCV: x right, y down, z forward) to the robot base frame. Constant over all frames and episodes of this dataset.",
 "cam1": {
  "fx_fy_cx_cy": [
   597.2525634765625,
   597.4514770507812,
   328.7528076171875,
   244.22544860839844
  ],
  "T_base_cam": [
   [
    0.9603833556175232,
    -0.16462299227714539,
    0.22486238181591034,
    0.18235638737678528
   ],
   [
    -0.2771751284599304,
    -0.4804461598396301,
    0.8320729732513428,
    -0.7760321497917175
   ],
   [
    -0.028944076970219612,
    -0.8614352941513062,
    -0.507041871547699,
    0.47138991951942444
   ],
   [
    0.0,
    0.0,
    0.0,
    1.0
   ]
  ]
 },
 "cam2": {
  "fx_fy_cx_cy": [
   601.3878173828125,
   601.7159423828125,
   333.2942199707031,
   245.56826782226562
  ],
  "T_base_cam": [
   [
    -0.9617871642112732,
    0.2029283493757248,
    -0.18380846083164215,
    0.5361539125442505
   ],
   [
    0.2541901469230652,
    0.4123094379901886,
    -0.8748647570610046,
    0.6881464719772339
   ],
   [
    -0.10174888372421265,
    -0.8881559371948242,
    -0.4481362998485565,
    0.4329851269721985
   ],
   [
    0.0,
    0.0,
    0.0,
    1.0
   ]
  ]
 }
}