def _get_proposal_pairs(self, proposals):
proposal_pairs = []
for i, proposals_per_image in enumerate(proposals):
box_subj = proposals_per_image.bbox
box_obj = proposals_per_image.bbox
box_subj = box_subj.unsqueeze(1).repeat(1, box_subj.shape[0], 1)
box_obj = box_obj.unsqueeze(0).repeat(box_obj.shape[0], 1, 1)
proposal_box_pairs = torch.cat((box_subj.view(-1, 4), box_obj.view(-1, 4)), 1)
idx_subj = torch.arange(box_subj.shape[0]).view(-1, 1, 1).repeat(1, box_obj.shape[0], 1).to(proposals_per_image.bbox.device)
idx_obj = torch.arange(box_obj.shape[0]).view(1, -1, 1).repeat(box_subj.shape[0], 1, 1).to(proposals_per_image.bbox.device)
proposal_idx_pairs = torch.cat((idx_subj.view(-1, 1), idx_obj.view(-1, 1)), 1)
keep_idx = (proposal_idx_pairs[:, 0] != proposal_idx_pairs[:, 1]).nonzero().view(-1)
# if we filter non overlap bounding boxes
if self.cfg.MODEL.ROI_RELATION_HEAD.FILTER_NON_OVERLAP:
ious = boxlist_iou(proposals_per_image, proposals_per_image).view(-1)
ious = ious[keep_idx]
keep_idx = keep_idx[(ious > 0).nonzero().view(-1)]
proposal_idx_pairs = proposal_idx_pairs[keep_idx]
proposal_box_pairs = proposal_box_pairs[keep_idx]
proposal_pairs_per_image = BoxPairList(proposal_box_pairs, proposals_per_image.size, proposals_per_image.mode)
proposal_pairs_per_image.add_field("idx_pairs", proposal_idx_pairs)
proposal_pairs.append(proposal_pairs_per_image)
return proposal_pairs
def match_targets_to_proposals(self, proposal, target):
match_quality_matrix = boxlist_iou(target, proposal)
temp = []
target_box_pairs = []
for i in range(match_quality_matrix.shape[0]):
for j in range(match_quality_matrix.shape[0]):
match_i = match_quality_matrix[i].view(1, -1)
match_j = match_quality_matrix[j].view(-1, 1)
match_ij = (match_i + match_j) / 2
match_ij.view(-1)[::match_quality_matrix.shape[1]] = 0
temp.append(match_ij)
boxi = target.bbox[i]
boxj = target.bbox[j]
box_pair = torch.cat((boxi, boxj), 0)
target_box_pairs.append(box_pair)
match_pair_quality_matrix = torch.stack(temp, 0).view(len(temp), -1)
target_box_pairs = torch.stack(target_box_pairs, 0)
target_pair = BoxPairList(target_box_pairs, target.size, target.mode)
target_pair.add_field("labels",
target.get_field("pred_labels").view(-1))
box_subj = proposal.bbox
box_obj = proposal.bbox
box_subj = box_subj.unsqueeze(1).repeat(1, box_subj.shape[0], 1)
box_obj = box_obj.unsqueeze(0).repeat(box_obj.shape[0], 1, 1)
proposal_box_pairs = torch.cat(
(box_subj.view(-1, 4), box_obj.view(-1, 4)), 1)
proposal_pairs = BoxPairList(proposal_box_pairs, proposal.size,
proposal.mode)
idx_subj = torch.arange(box_subj.shape[0]).view(-1, 1, 1).repeat(
1, box_obj.shape[0], 1).to(proposal.bbox.device)
idx_obj = torch.arange(box_obj.shape[0]).view(1, -1, 1).repeat(
box_subj.shape[0], 1, 1).to(proposal.bbox.device)
proposal_idx_pairs = torch.cat(
(idx_subj.view(-1, 1), idx_obj.view(-1, 1)), 1)
proposal_pairs.add_field("idx_pairs", proposal_idx_pairs)
# matched_idxs = self.proposal_matcher(match_quality_matrix)
matched_idxs = self.proposal_pair_matcher(match_pair_quality_matrix)
# Fast RCNN only need "labels" field for selecting the targets
# target = target.copy_with_fields("pred_labels")
# get the targets corresponding GT for each proposal
# NB: need to clamp the indices because we can have a single
# GT in the image, and matched_idxs can be -2, which goes
# out of bounds
if self.use_matched_pairs_only and (
matched_idxs >= 0).sum() > self.minimal_matched_pairs:
# filter all matched_idxs < 0
proposal_pairs = proposal_pairs[matched_idxs >= 0]
matched_idxs = matched_idxs[matched_idxs >= 0]
matched_targets = target_pair[matched_idxs.clamp(min=0)]
matched_targets.add_field("matched_idxs", matched_idxs)
return matched_targets, proposal_pairs
def match_targets_to_proposals(self, proposal, target):
match_quality_matrix = boxlist_iou(target, proposal)
matched_idxs = self.proposal_matcher(match_quality_matrix)
# Fast RCNN only need "labels" field for selecting the targets
target = target.copy_with_fields("labels")
# get the targets corresponding GT for each proposal
# NB: need to clamp the indices because we can have a single
# GT in the image, and matched_idxs can be -2, which goes
# out of bounds
matched_targets = target[matched_idxs.clamp(min=0)]
matched_targets.add_field("matched_idxs", matched_idxs)
return matched_targets
def match_targets_to_anchors(self, anchor, target, copied_fields=[]):
match_quality_matrix = boxlist_iou(target, anchor)
matched_idxs = self.proposal_matcher(match_quality_matrix)
# RPN doesn't need any fields from target
# for creating the labels, so clear them all
target = target.copy_with_fields(copied_fields)
# get the targets corresponding GT for each anchor
# NB: need to clamp the indices because we can have a single
# GT in the image, and matched_idxs can be -2, which goes
# out of bounds
matched_targets = target[matched_idxs.clamp(min=0)]
matched_targets.add_field("matched_idxs", matched_idxs)
return matched_targets
def _fullsample_test(self, proposals):
"""
This method get all subject-object pairs, and return the proposals.
Note: this function keeps a state.
Arguments:
proposals (list[BoxList])
"""
proposal_pairs = []
for i, proposals_per_image in enumerate(proposals):
box_subj = proposals_per_image.bbox
box_obj = proposals_per_image.bbox
box_subj = box_subj.unsqueeze(1).repeat(1, box_subj.shape[0], 1)
box_obj = box_obj.unsqueeze(0).repeat(box_obj.shape[0], 1, 1)
proposal_box_pairs = torch.cat(
(box_subj.view(-1, 4), box_obj.view(-1, 4)), 1)
idx_subj = torch.arange(box_subj.shape[0]).view(-1, 1, 1).repeat(
1, box_obj.shape[0], 1).to(proposals_per_image.bbox.device)
idx_obj = torch.arange(box_obj.shape[0]).view(1, -1, 1).repeat(
box_subj.shape[0], 1, 1).to(proposals_per_image.bbox.device)
proposal_idx_pairs = torch.cat(
(idx_subj.view(-1, 1), idx_obj.view(-1, 1)), 1)
keep_idx = (proposal_idx_pairs[:, 0] !=
proposal_idx_pairs[:, 1]).nonzero().view(-1)
# if we filter non overlap bounding boxes
if self.cfg.MODEL.ROI_RELATION_HEAD.FILTER_NON_OVERLAP:
ious = boxlist_iou(proposals_per_image,
proposals_per_image).view(-1)
ious = ious[keep_idx]
keep_idx = keep_idx[(ious > 0).nonzero().view(-1)]
proposal_idx_pairs = proposal_idx_pairs[keep_idx]
proposal_box_pairs = proposal_box_pairs[keep_idx]
proposal_pairs_per_image = BoxPairList(proposal_box_pairs,
proposals_per_image.size,
proposals_per_image.mode)
proposal_pairs_per_image.add_field("idx_pairs", proposal_idx_pairs)
proposal_pairs.append(proposal_pairs_per_image)
return proposal_pairs
def evaluate_box_proposals(predictions,
dataset,
thresholds=None,
area="all",
limit=None):
"""Evaluate detection proposal recall metrics. This function is a much
faster alternative to the official COCO API recall evaluation code. However,
it produces slightly different results.
"""
# Record max overlap value for each gt box
# Return vector of overlap values
areas = {
"all": 0,
"small": 1,
"medium": 2,
"large": 3,
"96-128": 4,
"128-256": 5,
"256-512": 6,
"512-inf": 7,
}
area_ranges = [
[0**2, 1e5**2], # all
[0**2, 32**2], # small
[32**2, 96**2], # medium
[96**2, 1e5**2], # large
[96**2, 128**2], # 96-128
[128**2, 256**2], # 128-256
[256**2, 512**2], # 256-512
[512**2, 1e5**2],
] # 512-inf
assert area in areas, "Unknown area range: {}".format(area)
area_range = area_ranges[areas[area]]
gt_overlaps = []
num_pos = 0
for image_id, prediction in enumerate(predictions):
original_id = image_id # dataset.id_to_img_map[image_id]
img_info = dataset.get_img_info(image_id)
image_width = img_info["width"]
image_height = img_info["height"]
prediction = prediction.resize((image_width, image_height))
# sort predictions in descending order
# TODO maybe remove this and make it explicit in the documentation
inds = prediction.get_field("objectness").sort(descending=True)[1]
prediction = prediction[inds]
ann_ids = dataset.coco.getAnnIds(imgIds=original_id)
anno = dataset.coco.loadAnns(ann_ids)
gt_boxes = [obj["bbox"] for obj in anno if obj["iscrowd"] == 0]
gt_boxes = torch.as_tensor(gt_boxes).reshape(
-1, 4) # guard against no boxes
gt_boxes = BoxList(gt_boxes, (image_width, image_height),
mode="xywh").convert("xyxy")
gt_areas = torch.as_tensor(
[obj["area"] for obj in anno if obj["iscrowd"] == 0])
if len(gt_boxes) == 0:
continue
valid_gt_inds = (gt_areas >= area_range[0]) & (gt_areas <=
area_range[1])
gt_boxes = gt_boxes[valid_gt_inds]
num_pos += len(gt_boxes)
if len(gt_boxes) == 0:
continue
if len(prediction) == 0:
continue
if limit is not None and len(prediction) > limit:
prediction = prediction[:limit]
overlaps = boxlist_iou(prediction, gt_boxes)
_gt_overlaps = torch.zeros(len(gt_boxes))
for j in range(min(len(prediction), len(gt_boxes))):
# find which proposal box maximally covers each gt box
# and get the iou amount of coverage for each gt box
max_overlaps, argmax_overlaps = overlaps.max(dim=0)
# find which gt box is 'best' covered (i.e. 'best' = most iou)
gt_ovr, gt_ind = max_overlaps.max(dim=0)
assert gt_ovr >= 0
# find the proposal box that covers the best covered gt box
box_ind = argmax_overlaps[gt_ind]
# record the iou coverage of this gt box
_gt_overlaps[j] = overlaps[box_ind, gt_ind]
assert _gt_overlaps[j] == gt_ovr
# mark the proposal box and the gt box as used
overlaps[box_ind, :] = -1
overlaps[:, gt_ind] = -1
# append recorded iou coverage level
gt_overlaps.append(_gt_overlaps)
gt_overlaps = torch.cat(gt_overlaps, dim=0)
gt_overlaps, _ = torch.sort(gt_overlaps)
if thresholds is None:
step = 0.05
thresholds = torch.arange(0.5, 0.95 + 1e-5, step, dtype=torch.float32)
recalls = torch.zeros_like(thresholds)
# compute recall for each iou threshold
for i, t in enumerate(thresholds):
recalls[i] = (gt_overlaps >= t).float().sum() / float(num_pos)
# ar = 2 * np.trapz(recalls, thresholds)
ar = recalls.mean()
return {
"ar": ar,
"recalls": recalls,
"thresholds": thresholds,
"gt_overlaps": gt_overlaps,
"num_pos": num_pos,
}
def calc_detection_voc_prec_rec(gt_boxlists, pred_boxlists, iou_thresh=0.5):
"""Calculate precision and recall based on evaluation code of PASCAL VOC.
This function calculates precision and recall of
predicted bounding boxes obtained from a dataset which has :math:`N`
images.
The code is based on the evaluation code used in PASCAL VOC Challenge.
"""
n_pos = defaultdict(int)
score = defaultdict(list)
match = defaultdict(list)
for gt_boxlist, pred_boxlist in zip(gt_boxlists, pred_boxlists):
pred_bbox = pred_boxlist.bbox.numpy()
pred_label = pred_boxlist.get_field("labels").numpy()
pred_score = pred_boxlist.get_field("scores").numpy()
gt_bbox = gt_boxlist.bbox.numpy()
gt_label = gt_boxlist.get_field("labels").numpy()
gt_difficult = gt_boxlist.get_field("difficult").numpy()
for l in np.unique(np.concatenate((pred_label, gt_label)).astype(int)):
pred_mask_l = pred_label == l
pred_bbox_l = pred_bbox[pred_mask_l]
pred_score_l = pred_score[pred_mask_l]
# sort by score
order = pred_score_l.argsort()[::-1]
pred_bbox_l = pred_bbox_l[order]
pred_score_l = pred_score_l[order]
gt_mask_l = gt_label == l
gt_bbox_l = gt_bbox[gt_mask_l]
gt_difficult_l = gt_difficult[gt_mask_l]
n_pos[l] += np.logical_not(gt_difficult_l).sum()
score[l].extend(pred_score_l)
if len(pred_bbox_l) == 0:
continue
if len(gt_bbox_l) == 0:
match[l].extend((0, ) * pred_bbox_l.shape[0])
continue
# VOC evaluation follows integer typed bounding boxes.
pred_bbox_l = pred_bbox_l.copy()
pred_bbox_l[:, 2:] += 1
gt_bbox_l = gt_bbox_l.copy()
gt_bbox_l[:, 2:] += 1
iou = boxlist_iou(
BoxList(pred_bbox_l, gt_boxlist.size),
BoxList(gt_bbox_l, gt_boxlist.size),
).numpy()
gt_index = iou.argmax(axis=1)
# set -1 if there is no matching ground truth
gt_index[iou.max(axis=1) < iou_thresh] = -1
del iou
selec = np.zeros(gt_bbox_l.shape[0], dtype=bool)
for gt_idx in gt_index:
if gt_idx >= 0:
if gt_difficult_l[gt_idx]:
match[l].append(-1)
else:
if not selec[gt_idx]:
match[l].append(1)
else:
match[l].append(0)
selec[gt_idx] = True
else:
match[l].append(0)
n_fg_class = max(n_pos.keys()) + 1
prec = [None] * n_fg_class
rec = [None] * n_fg_class
for l in n_pos.keys():
score_l = np.array(score[l])
match_l = np.array(match[l], dtype=np.int8)
order = score_l.argsort()[::-1]
match_l = match_l[order]
tp = np.cumsum(match_l == 1)
fp = np.cumsum(match_l == 0)
# If an element of fp + tp is 0,
# the corresponding element of prec[l] is nan.
prec[l] = tp / (fp + tp)
# If n_pos[l] is 0, rec[l] is None.
if n_pos[l] > 0:
rec[l] = tp / n_pos[l]
return prec, rec
def _relpnsample_test(self, proposals):
"""
perform relpn based sampling during testing
"""
proposals[0] = proposals[0]
proposal_pairs = self._fullsample_test(proposals)
proposal_pairs = list(proposal_pairs)
relnesses = []
for img_idx, proposals_per_image in enumerate(proposals):
obj_logits = proposals_per_image.get_field('logits')
obj_bboxes = proposals_per_image.bbox
relness = self.relationshipness(obj_logits, obj_bboxes,
proposals_per_image.size)
keep_idx = (1 - torch.eye(obj_logits.shape[0]).to(
relness.device)).view(-1).nonzero().view(-1)
if self.cfg.MODEL.ROI_RELATION_HEAD.FILTER_NON_OVERLAP:
ious = boxlist_iou(proposals_per_image,
proposals_per_image).view(-1)
ious = ious[keep_idx]
keep_idx = keep_idx[(ious > 0).nonzero().view(-1)]
relness = relness.view(-1)[keep_idx]
relness_sorted, order = torch.sort(relness.view(-1),
descending=True)
# perform co-nms to filter duplicate bounding boxes
# ious = boxlist_iou(proposals_per_image, proposals_per_image)
# subj_ids = []; obj_ids = []
# sample_ids = []; id = 0
# while len(sample_ids) < self.cfg.MODEL.ROI_RELATION_HEAD.BATCH_SIZE_PER_IMAGE and id < len(order):
# subj_id = order[id] / len(proposals_per_image)
# obj_id = order[id] % len(proposals_per_image)
#
# if len(subj_ids) == 0 and len(obj_ids) == 0 and subj_id != obj_id:
# subj_ids.append(subj_id.item())
# obj_ids.append(obj_id.item())
# sample_ids.append(id)
# else:
# subj_ious = ious[subj_id, subj_ids]
# obj_ious = ious[obj_id, obj_ids]
# if (subj_ious.max() < 0.9 or obj_ious.max() < 0.9) and subj_id != obj_id:
# subj_ids.append(subj_id.item())
# obj_ids.append(obj_id.item())
# sample_ids.append(id)
# id += 1
# img_sampled_inds = order[sample_ids]
# relness = relness_sorted[sample_ids]
img_sampled_inds = order[:self.cfg.MODEL.ROI_RELATION_HEAD.
BATCH_SIZE_PER_IMAGE].view(-1)
relness = relness_sorted[:self.cfg.MODEL.ROI_RELATION_HEAD.
BATCH_SIZE_PER_IMAGE].view(-1)
proposal_pairs_per_image = proposal_pairs[img_idx][
img_sampled_inds]
proposal_pairs[img_idx] = proposal_pairs_per_image
relnesses.append(relness)
self._proposal_pairs = proposal_pairs
return proposal_pairs, relnesses
