Spaces:
Runtime error
Runtime error
File size: 7,400 Bytes
4ea50ff |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 |
# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
import math
import torch
import pandas as pd
from maskrcnn_benchmark.data.datasets.evaluation.word import io_
class BoxCoder(object):
"""
This class encodes and decodes a set of bounding boxes into the representation used for training the regressors.
"""
def __init__(self, weights, bbox_xform_clip=math.log(1000. / 16)):
"""
Arguments:
weights (4-element tuple)
bbox_xform_clip (float)
"""
self.weights = weights
self.bbox_xform_clip = bbox_xform_clip
def encode(self, reference_boxes, proposals):
"""
Encode a set of proposals with respect to some
reference boxes
Arguments:
reference_boxes (Tensor): reference boxes
proposals (Tensor): boxes to be encoded
"""
TO_REMOVE = 1 # TODO remove
ex_widths = proposals[:, 2] - proposals[:, 0] + TO_REMOVE
ex_heights = proposals[:, 3] - proposals[:, 1] + TO_REMOVE
ex_ctr_x = proposals[:, 0] + 0.5 * ex_widths
ex_ctr_y = proposals[:, 1] + 0.5 * ex_heights
gt_widths = reference_boxes[:, 2] - reference_boxes[:, 0] + TO_REMOVE
gt_heights = reference_boxes[:, 3] - reference_boxes[:, 1] + TO_REMOVE
gt_ctr_x = reference_boxes[:, 0] + 0.5 * gt_widths
gt_ctr_y = reference_boxes[:, 1] + 0.5 * gt_heights
wx, wy, ww, wh = self.weights
targets_dx = wx * (gt_ctr_x - ex_ctr_x) / ex_widths
targets_dy = wy * (gt_ctr_y - ex_ctr_y) / ex_heights
targets_dw = ww * torch.log(gt_widths / ex_widths)
targets_dh = wh * torch.log(gt_heights / ex_heights)
targets = torch.stack((targets_dx, targets_dy, targets_dw, targets_dh), dim=1)
return targets
def encode_iou(self, reference_boxes, proposals):
"""
Encode a set of proposals with respect to some
reference boxes
Arguments:
reference_boxes (Tensor): reference boxes
proposals (Tensor): boxes to be encoded
"""
TO_REMOVE = 1 # TODO remove
ex_widths = proposals[:, 2] - proposals[:, 0] + TO_REMOVE
ex_heights = proposals[:, 3] - proposals[:, 1] + TO_REMOVE
ex_ctr_x = proposals[:, 0] + 0.5 * ex_widths
ex_ctr_y = proposals[:, 1] + 0.5 * ex_heights
gt_widths = reference_boxes[:, 2] - reference_boxes[:, 0] + TO_REMOVE
gt_heights = reference_boxes[:, 3] - reference_boxes[:, 1] + TO_REMOVE
gt_ctr_x = reference_boxes[:, 0] + 0.5 * gt_widths
gt_ctr_y = reference_boxes[:, 1] + 0.5 * gt_heights
wx, wy, ww, wh = self.weights
targets_dx = wx * (gt_ctr_x - ex_ctr_x) / ex_widths
targets_dy = wy * (gt_ctr_y - ex_ctr_y) / ex_heights
targets_dw = ww * torch.log(gt_widths / ex_widths)
targets_dh = wh * torch.log(gt_heights / ex_heights)
targets = torch.stack((targets_dx, targets_dy, targets_dw, targets_dh), dim=1)
return targets
def decode(self, rel_codes, boxes):
"""
From a set of original boxes and encoded relative box offsets,
get the decoded boxes.
Arguments:
rel_codes (Tensor): encoded boxes # predict [2, 12000, 4]
boxes (Tensor): reference boxes. # anchor [2, 12000, 4] xmin0 ymin1 xmax2 ymax3
"""
boxes = boxes.to(rel_codes.dtype)
TO_REMOVE = 1 # TODO remove
widths = boxes[:, 2] - boxes[:, 0] + TO_REMOVE
heights = boxes[:, 3] - boxes[:, 1] + TO_REMOVE
ctr_x = boxes[:, 0] + 0.5 * widths
ctr_y = boxes[:, 1] + 0.5 * heights
wx, wy, ww, wh = self.weights
dx = rel_codes[:, 0::4] / wx
dy = rel_codes[:, 1::4] / wy
dw = rel_codes[:, 2::4] / ww
dh = rel_codes[:, 3::4] / wh
dw = torch.clamp(dw, max=self.bbox_xform_clip)
dh = torch.clamp(dh, max=self.bbox_xform_clip)
pred_ctr_x = dx * widths[:, None] + ctr_x[:, None]
pred_ctr_y = dy * heights[:, None] + ctr_y[:, None]
pred_w = torch.exp(dw) * widths[:, None]
pred_h = torch.exp(dh) * heights[:, None]
##############################
pred_boxes = torch.zeros_like(rel_codes)
pred_boxes[:, 0::4] = pred_ctr_x - 0.5 * pred_w
pred_boxes[:, 1::4] = pred_ctr_y - 0.5 * pred_h
pred_boxes[:, 2::4] = pred_ctr_x + 0.5 * pred_w - 1
pred_boxes[:, 3::4] = pred_ctr_y + 0.5 * pred_h - 1
return pred_boxes
def decode_iou(self, rel_codes, boxes, num_p = 8):
"""
From a set of original boxes and encoded relative box offsets,
get the decoded boxes.
Arguments:
rel_codes (Tensor): encoded boxes # predict [2, 12000, 4]
boxes (Tensor): reference boxes. # anchor [2, 12000, 4] xmin0 ymin1 xmax2 ymax3
"""
boxes = boxes.to(rel_codes.dtype)
TO_REMOVE = 1 # TODO remove
widths = boxes[:, 2] - boxes[:, 0] + TO_REMOVE
heights = boxes[:, 3] - boxes[:, 1] + TO_REMOVE
ctr_x = boxes[:, 0] + 0.5 * widths
ctr_y = boxes[:, 1] + 0.5 * heights
# 123
# 8#4
# 765
if num_p == 8: # 8 boundary points
x_1 = boxes[:, 0] + widths * rel_codes[:, 0]
y_1 = boxes[:, 1] + heights * rel_codes[:, 1]
x_2 = ctr_x + widths * rel_codes[:, 2]
y_2 = boxes[:, 1] + heights * rel_codes[:, 3]
x_3 = boxes[:, 2] + widths * rel_codes[:, 4]
y_3 = boxes[:, 1] + heights * rel_codes[:, 5]
x_4 = boxes[:, 2] + widths * rel_codes[:, 6]
y_4 = ctr_y + heights * rel_codes[:, 7]
x_5 = boxes[:, 2] + widths * rel_codes[:, 8]
y_5 = boxes[:, 3] + heights * rel_codes[:, 9]
x_6 = ctr_x + widths * rel_codes[:, 10]
y_6 = boxes[:, 3] + heights * rel_codes[:, 11]
x_7 = boxes[:, 0] + widths * rel_codes[:, 12]
y_7 = boxes[:, 3] + heights * rel_codes[:, 13]
x_8 = boxes[:, 0] + widths * rel_codes[:, 14]
y_8 = ctr_y + heights * rel_codes[:, 15]
x_total = torch.stack([x_1, x_2, x_3, x_4, x_5, x_6, x_7, x_8], 0)
y_total = torch.stack([y_1, y_2, y_3, y_4, y_5, y_6, y_7, y_8], 0)
x_min = torch.min(x_total, 0, keepdim=True) # [1, N]
x_max = torch.max(x_total, 0, keepdim=True)
y_min = torch.min(y_total, 0, keepdim=True)
y_max = torch.max(y_total, 0, keepdim=True)
N1, N2 = x_min[0].shape
x_min = x_min[0].view([N2])
x_max = x_max[0].view([N2])
y_min = y_min[0].view([N2])
y_max = y_max[0].view([N2])
x_min = torch.stack([x_min, ctr_x], 0)
x_max = torch.stack([x_max, ctr_x], 0)
y_min = torch.stack([y_min, ctr_y], 0)
y_max = torch.stack([y_max, ctr_y], 0)
x_min = torch.min(x_min, 0, keepdim=True) # [1, N]
x_max = torch.max(x_max, 0, keepdim=True)
y_min = torch.min(y_min, 0, keepdim=True)
y_max = torch.max(y_max, 0, keepdim=True)
pred_boxes = torch.zeros_like(boxes)
pred_boxes[:, 0] = x_min[0][0, :]
pred_boxes[:, 1] = y_min[0][0, :]
pred_boxes[:, 2] = x_max[0][0, :]
pred_boxes[:, 3] = y_max[0][0, :]
return pred_boxes
|