Extracts:
image: 000001.jpg,...
depth: 000001.png,...
label:[classname, centroid1,centroid2,centroid3,w,l,h,orientation1,orientatio2]
pcd: [N,6] point cloud data created from depth image
np.array([N, 6]) N = number of points
pcd = np.load("pcd/000001.npz")
points = pcd[:, 0:3] , colors = pcd[:, 3:6]
calib: {Rtilt: [[x,x,x],[x,x,x],[x,x,x]], K:[[fx,0,cx],[0,fy,cy],[0,0,1]]}
with open("calib/000001.txt", "r") as file:
calib = json.load(file)
Rtilt = calib["Rtilt]
K = calib["K"]
votes: np.load("votes/000001.npz")
VISUALIZE PCD:
VISUALIZE 3D BBOXES ON 2D IMAGE:
