Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

【提问】关于使用PaddlePaddle SMOKE模型进行对象检测时的cam_info参数的疑惑 #5539

Open
Creling opened this issue Jul 22, 2022 · 1 comment
Assignees

Comments

@Creling
Copy link

Creling commented Jul 22, 2022

大家好:

如题,在使用PaddlePaddle SMOKE进行物体识别时,除了需要传入img外,还需要传入一个cam_info参数,从代码中可以看到,cam_info包含了一个硬编码的矩阵K

K = np.array([[[2055.56, 0, 939.658], [0, 2055.56, 641.072], [0, 0, 1]]], np.float32)
K_inverse = np.linalg.inv(K)
K_inverse = paddle.to_tensor(K_inverse)
img, ori_img_size, output_size = get_img(args.input_path)
ratio = get_ratio(ori_img_size, output_size)
ratio = paddle.to_tensor(ratio)
cam_info = [K_inverse, ratio]
total_pred = model(img, cam_info)

我的疑惑是,这个K矩阵是如何计算出的?使用SMOKE的官方实现进行物体识别时并不需要传入这样的一个矩阵。

目前,我猜测这是为了兼容不同数据集而引入的一个特定于数据集的坐标转换矩阵,但从维度和数值来看,和训练PaddlePaddle SMOKE时使用的KITTI的数值标定矩阵似乎也并不相符?

请教 @michaelowenliu :)

@mikasazgx
Copy link

相机的内参矩阵

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants