Files
pytorch/caffe2/operators/bbox_transform_op.h
Mingzhe Li 964e30de1d Workaround for Cuda9.2 and GCC7 compilation errors (#10510)
Summary:
Breaking out of #8338

This PR is a workaround for a bug with CUDA9.2 + GCC7.

Here is the error this PR fixed:
.../pytorch/caffe2/operators/elementwise_ops.h: In constructor ‘caffe2::BinaryElementwiseWithArgsOp<InputTypes, Context, Functor, OutputTypeMap>::BinaryElementwiseWithArgsOp(const caffe2::OperatorDef&, caffe2::Workspace*)’:
.../pytorch/caffe2/operators/elementwise_ops.h:106:189: error: ‘GetSingleArgument<bool>’ is not a member of ‘caffe2::BinaryElementwiseWithArgsOp<InputTypes, Context, Functor, OutputTypeMap>’
   BinaryElementwiseWithArgsOp(const OperatorDef& operator_def, Workspace* ws)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/10510

Reviewed By: orionr

Differential Revision: D9319742

Pulled By: mingzhe09088

fbshipit-source-id: ce59e3db14539f071f3c20301e77ca36a6fc3f81
2018-08-14 20:54:52 -07:00

73 lines
2.6 KiB
C++

// Copyright 2004-present Facebook. All Rights Reserved.
#ifndef BBOX_TRANSFORM_OP_H_
#define BBOX_TRANSFORM_OP_H_
#include "caffe2/core/context.h"
#include "caffe2/core/logging.h"
#include "caffe2/core/operator.h"
#include "caffe2/utils/math.h"
namespace caffe2 {
template <typename T, class Context>
class BBoxTransformOp final : public Operator<Context> {
public:
BBoxTransformOp(const OperatorDef& operator_def, Workspace* ws)
: Operator<Context>(operator_def, ws),
weights_(this->template GetRepeatedArgument<T>(
"weights",
vector<T>{1.0f, 1.0f, 1.0f, 1.0f})),
apply_scale_(
this->template GetSingleArgument<bool>("apply_scale", true)),
correct_transform_coords_(this->template GetSingleArgument<bool>(
"correct_transform_coords",
false)),
rotated_(this->template GetSingleArgument<bool>("rotated", false)),
angle_bound_on_(
this->template GetSingleArgument<bool>("angle_bound_on", true)),
angle_bound_lo_(
this->template GetSingleArgument<int>("angle_bound_lo", -90)),
angle_bound_hi_(
this->template GetSingleArgument<int>("angle_bound_hi", 90)),
clip_angle_thresh_(
this->template GetSingleArgument<float>("clip_angle_thresh", 1.0)) {
CAFFE_ENFORCE_EQ(
weights_.size(),
4,
"weights size " + caffe2::to_string(weights_.size()) + "must be 4.");
}
USE_OPERATOR_CONTEXT_FUNCTIONS;
bool RunOnDevice() override;
protected:
// weights [wx, wy, ww, wh] to apply to the regression target
vector<T> weights_;
// Transform the boxes to the scaled image space after applying the bbox
// deltas.
// Set to false to match the detectron code, set to true for the keypoint
// model and for backward compatibility
bool apply_scale_{true};
// Correct bounding box transform coordates, see bbox_transform() in boxes.py
// Set to true to match the detectron code, set to false for backward
// compatibility
bool correct_transform_coords_{false};
// Set for RRPN case to handle rotated boxes. Inputs should be in format
// [ctr_x, ctr_y, width, height, angle (in degrees)].
bool rotated_{false};
// If set, for rotated boxes in RRPN, output angles are normalized to be
// within [angle_bound_lo, angle_bound_hi].
bool angle_bound_on_{true};
int angle_bound_lo_{-90};
int angle_bound_hi_{90};
// For RRPN, clip almost horizontal boxes within this threshold of
// tolerance for backward compatibility. Set to negative value for
// no clipping.
float clip_angle_thresh_{1.0};
};
} // namespace caffe2
#endif // BBOX_TRANSFORM_OP_H_