search

rhett-chen / graspness_implementation

My implementation of Graspnet Graspness.
Other
139 stars 33 forks source link

Intrinsic Camera #32

Closed LeVHoangduc closed 5 months ago

LeVHoangduc commented 5 months ago

I want to check image from camera image. I use L515 realsense to do it. But I have some question about generate cloud from depth image.

At function create_point_cloud_from_depth_image(depth, camera, organized=True), it has a camera parameter. Camera is generated by
camera = CameraInfo(1280.0, 720.0, intrinsic[0][0], intrinsic[1][1], intrinsic[0][2], intrinsic[1][2], factor_depth) So, I can get a width, height, fx,fy,cx,cy of Camera. But I'm confused with factor_depth? I tried to print() depth value from dataset, it's [[1000.]], why is number? For my case, I get a depth value that means depth scale is 0.0002500000118743628! I read this document and the scale value is also 1000.0. link So, Could you help me to clarify this number?

rhett-chen commented 5 months ago

In the GraspNet-1Billion dataset, the actual depth values are multiplied by 1000 to convert them into integers, which then allows the depth values to be stored as an image. Hence, the factor_depth is 1000. If your get the actual depth values, then you can simply set the factor_depth to 1.

LeVHoangduc commented 5 months ago

@rhett-chen Sorry for the late reply. I got it. Thanks a lot for your support