Theta(angle) in rbox

[d5423197]

yolov5_obb/utils/loss.py

Lines 170 to 172 in e4dfa87

	# theta Classification by Circular Smooth Label
	t_theta = tgaussian_theta[i].type(ps.dtype) # target theta_gaussian_labels
	ltheta += self.BCEtheta(ps[:, class_index:], t_theta)

作者你好,

感谢您的开源, 在训练时和在预测阶段, 这个rbox中的angle都没有被用到好像, 是这样的吗?

谢谢,

ZD