mm系列包hfjehfkevmkd
立即下载
资源介绍:
mm系列包hfjehfkevmkd
#include "pytorch_cpp_helper.hpp"
#include "pytorch_device_registry.hpp"
void AssignScoreWithKForwardCUDAKernelLauncher(
int B, int N0, int N1, int M, int K, int O, int aggregate,
const Tensor& points, const Tensor& centers, const Tensor& scores,
const Tensor& knn_idx, Tensor& output);
void AssignScoreWithKBackwardCUDAKernelLauncher(
int B, int N0, int N1, int M, int K, int O, int aggregate,
const Tensor& grad_out, const Tensor& points, const Tensor& centers,
const Tensor& scores, const Tensor& knn_idx, Tensor& grad_points,
Tensor& grad_centers, Tensor& grad_scores);
void assign_score_withk_forward_cuda(int B, int N0, int N1, int M, int K, int O,
int aggregate, const Tensor& points,
const Tensor& centers,
const Tensor& scores,
const Tensor& knn_idx, Tensor& output) {
AssignScoreWithKForwardCUDAKernelLauncher(
B, N0, N1, M, K, O, aggregate, points, centers, scores, knn_idx, output);
};
void assign_score_withk_backward_cuda(
int B, int N0, int N1, int M, int K, int O, int aggregate,
const Tensor& grad_out, const Tensor& points, const Tensor& centers,
const Tensor& scores, const Tensor& knn_idx, Tensor& grad_points,
Tensor& grad_centers, Tensor& grad_scores) {
AssignScoreWithKBackwardCUDAKernelLauncher(
B, N0, N1, M, K, O, aggregate, grad_out, points, centers, scores, knn_idx,
grad_points, grad_centers, grad_scores);
};
void assign_score_withk_forward_impl(int B, int N0, int N1, int M, int K, int O,
int aggregate, const Tensor& points,
const Tensor& centers,
const Tensor& scores,
const Tensor& knn_idx, Tensor& output);
void assign_score_withk_backward_impl(
int B, int N0, int N1, int M, int K, int O, int aggregate,
const Tensor& grad_out, const Tensor& points, const Tensor& centers,
const Tensor& scores, const Tensor& knn_idx, Tensor& grad_points,
Tensor& grad_centers, Tensor& grad_scores);
REGISTER_DEVICE_IMPL(assign_score_withk_forward_impl, CUDA,
assign_score_withk_forward_cuda);
REGISTER_DEVICE_IMPL(assign_score_withk_backward_impl, CUDA,
assign_score_withk_backward_cuda);
void BallQueryForwardCUDAKernelLauncher(int b, int n, int m, float min_radius,
float max_radius, int nsample,
const Tensor new_xyz, const Tensor xyz,
Tensor idx);
void ball_query_forward_cuda(int b, int n, int m, float min_radius,
float max_radius, int nsample,
const Tensor new_xyz, const Tensor xyz,
Tensor idx) {
BallQueryForwardCUDAKernelLauncher(b, n, m, min_radius, max_radius, nsample,
new_xyz, xyz, idx);
};
void ball_query_forward_impl(int b, int n, int m, float min_radius,
float max_radius, int nsample,
const Tensor new_xyz, const Tensor xyz,
Tensor idx);
REGISTER_DEVICE_IMPL(ball_query_forward_impl, CUDA, ball_query_forward_cuda);
void BBoxOverlapsCUDAKernelLauncher(const Tensor bboxes1, const Tensor bboxes2,
Tensor ious, const int mode,
const bool aligned, const int offset);
void bbox_overlaps_cuda(const Tensor bboxes1, const Tensor bboxes2, Tensor ious,
const int mode, const bool aligned, const int offset) {
BBoxOverlapsCUDAKernelLauncher(bboxes1, bboxes2, ious, mode, aligned, offset);
}
void bbox_overlaps_impl(const Tensor bboxes1, const Tensor bboxes2, Tensor ious,
const int mode, const bool aligned, const int offset);
REGISTER_DEVICE_IMPL(bbox_overlaps_impl, CUDA, bbox_overlaps_cuda);
void BorderAlignForwardCUDAKernelLauncher(const Tensor& input,
const Tensor& boxes, Tensor output,
Tensor argmax_idx,
const int pool_size);
void BorderAlignBackwardCUDAKernelLauncher(const Tensor& grad_output,
const Tensor& boxes,
const Tensor& argmax_idx,
Tensor grad_input,
const int pool_size);
void border_align_forward_cuda(const Tensor& input, const Tensor& boxes,
Tensor output, Tensor argmax_idx,
const int pool_size) {
BorderAlignForwardCUDAKernelLauncher(input, boxes, output, argmax_idx,
pool_size);
}
void border_align_backward_cuda(const Tensor& grad_output, const Tensor& boxes,
const Tensor& argmax_idx, Tensor grad_input,
const int pool_size) {
BorderAlignBackwardCUDAKernelLauncher(grad_output, boxes, argmax_idx,
grad_input, pool_size);
}
void border_align_forward_impl(const Tensor& input, const Tensor& boxes,
Tensor output, Tensor argmax_idx,
const int pool_size);
void border_align_backward_impl(const Tensor& grad_output, const Tensor& boxes,
const Tensor& argmax_idx, Tensor grad_input,
const int pool_size);
REGISTER_DEVICE_IMPL(border_align_forward_impl, CUDA,
border_align_forward_cuda);
REGISTER_DEVICE_IMPL(border_align_backward_impl, CUDA,
border_align_backward_cuda);
void box_iou_rotated_cuda(const Tensor boxes1, const Tensor boxes2, Tensor ious,
const int mode_flag, const bool aligned);
void box_iou_rotated_impl(const Tensor boxes1, const Tensor boxes2, Tensor ious,
const int mode_flag, const bool aligned);
REGISTER_DEVICE_IMPL(box_iou_rotated_impl, CUDA, box_iou_rotated_cuda);
void CARAFEForwardCUDAKernelLauncher(const Tensor features, const Tensor masks,
Tensor rfeatures, Tensor routput,
Tensor rmasks, Tensor output,
const int kernel_size,
const int group_size,
const int scale_factor);
void CARAFEBackwardCUDAKernelLauncher(
const Tensor top_grad, const Tensor rfeatures, const Tensor masks,
Tensor rtop_grad, Tensor rbottom_grad_hs, Tensor rbottom_grad,
Tensor rmask_grad, Tensor bottom_grad, Tensor mask_grad,
const int kernel_size, const int group_size, const int scale_factor);
void carafe_forward_cuda(Tensor features, Tensor masks, Tensor rfeatures,
Tensor routput, Tensor rmasks, Tensor output,
int kernel_size, int group_size, int scale_factor) {
CARAFEForwardCUDAKernelLauncher(features, masks, rfeatures, routput, rmasks,
output, kernel_size, group_size,
scale_factor);
}
void carafe_backward_cuda(Tensor top_grad, Tensor rfeatures, Tensor masks,
Tensor rtop_grad, Tensor rbottom_grad_hs,
Tensor rbottom_grad, Tensor rmask_grad,
Tensor bottom_grad, Tensor mask_grad, int kernel_size,
int group_size, int scale_factor) {
CARAFEBackwardCUDAKernelLauncher(top_grad, rfeatures, masks, rtop_grad,
rbottom_grad_hs, rbottom_grad, rmask_gr
资源文件列表:
mm.zip 大约有5768个文件