/
softgroup.py
313 lines (264 loc) · 15.1 KB
/
softgroup.py
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
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
import numpy as np
import torch.nn as nn
from minsu3d.evaluation.instance_segmentation import get_gt_instances, rle_encode
from minsu3d.evaluation.object_detection import get_gt_bbox
from minsu3d.common_ops.functions import softgroup_ops, common_ops
from minsu3d.evaluation.semantic_segmentation import *
from minsu3d.model.module import TinyUnet
from minsu3d.model.general_model import GeneralModel, clusters_voxelization
class SoftGroup(GeneralModel):
def __init__(self, cfg):
super().__init__(cfg)
output_channel = cfg.model.network.m
self.instance_classes = cfg.data.classes - len(cfg.data.ignore_classes)
"""
Top-down Refinement Block
"""
self.tiny_unet = TinyUnet(output_channel)
self.classification_branch = nn.Linear(output_channel, self.instance_classes + 1)
self.mask_scoring_branch = nn.Sequential(
nn.Linear(output_channel, output_channel),
nn.ReLU(inplace=True),
nn.Linear(output_channel, self.instance_classes + 1)
)
self.iou_score = nn.Linear(output_channel, self.instance_classes + 1)
def forward(self, data_dict):
output_dict = super().forward(data_dict)
if self.current_epoch > self.hparams.cfg.model.network.prepare_epochs:
"""
Top-down Refinement Block
"""
semantic_scores = output_dict["semantic_scores"].softmax(dim=-1)
proposals_offset_list = []
proposals_idx_list = []
for class_id in range(self.hparams.cfg.data.classes):
if class_id + 1 in self.hparams.cfg.data.ignore_classes:
continue
scores = semantic_scores[:, class_id].contiguous()
object_idxs = (scores > self.hparams.cfg.model.network.grouping_cfg.score_thr).nonzero().view(-1)
if object_idxs.size(0) < self.hparams.cfg.model.network.test_cfg.min_npoint:
continue
batch_idxs_ = data_dict["vert_batch_ids"][object_idxs]
batch_offsets_ = torch.cumsum(torch.bincount(batch_idxs_ + 1), dim=0).int()
coords_ = data_dict["point_xyz"][object_idxs]
pt_offsets_ = output_dict["point_offsets"][object_idxs]
idx, start_len = common_ops.ballquery_batch_p(
coords_ + pt_offsets_, batch_idxs_, batch_offsets_,
self.hparams.cfg.model.network.grouping_cfg.radius,
self.hparams.cfg.model.network.grouping_cfg.mean_active
)
proposals_idx, proposals_offset = softgroup_ops.sg_bfs_cluster(
self.hparams.cfg.data.point_num_avg, idx.cpu(),
start_len.cpu(),
self.hparams.cfg.model.network.grouping_cfg.npoint_thr, class_id)
proposals_idx = proposals_idx.long().to(self.device)
proposals_offset = proposals_offset.to(self.device)
proposals_idx[:, 1] = object_idxs[proposals_idx[:, 1]]
# merge proposals
if len(proposals_offset_list) > 0:
proposals_idx[:, 0] += sum([x.size(0) for x in proposals_offset_list]) - 1
proposals_offset += proposals_offset_list[-1][-1]
proposals_offset = proposals_offset[1:]
if proposals_idx.size(0) > 0:
proposals_idx_list.append(proposals_idx)
proposals_offset_list.append(proposals_offset)
proposals_idx = torch.cat(proposals_idx_list, dim=0)
proposals_offset = torch.cat(proposals_offset_list)
if proposals_offset.shape[0] > self.hparams.cfg.model.network.train_cfg.max_proposal_num:
proposals_offset = proposals_offset[:self.hparams.cfg.model.network.train_cfg.max_proposal_num + 1]
proposals_idx = proposals_idx[:proposals_offset[-1]]
assert proposals_idx.shape[0] == proposals_offset[-1]
output_dict["proposals_idx"] = proposals_idx
output_dict["proposals_offset"] = proposals_offset
inst_feats, inst_map = clusters_voxelization(
clusters_idx=proposals_idx,
clusters_offset=proposals_offset,
feats=output_dict["point_features"],
coords=data_dict["point_xyz"],
device=self.device,
**self.hparams.cfg.model.network.instance_voxel_cfg
)
feats = self.tiny_unet(inst_feats)
# predict mask scores
mask_scores = self.mask_scoring_branch(feats.features)
output_dict["mask_scores"] = mask_scores[inst_map]
output_dict["instance_batch_idxs"] = feats.coordinates[:, 0][inst_map]
# predict instance cls and iou scores
feats = self.global_pool(feats)
output_dict["cls_scores"] = self.classification_branch(feats)
output_dict["iou_scores"] = self.iou_score(feats)
return output_dict
def global_pool(self, x, expand=False):
indices = x.coordinates[:, 0]
batch_offset = torch.cumsum(torch.bincount(indices + 1), dim=0).int()
x_pool = softgroup_ops.global_avg_pool(x.features, batch_offset)
if not expand:
return x_pool
x_pool_expand = x_pool[indices.long()]
x.features = torch.cat((x.features, x_pool_expand), dim=1)
return x
def _loss(self, data_dict, output_dict):
losses = super()._loss(data_dict, output_dict)
if self.current_epoch > self.hparams.cfg.model.network.prepare_epochs:
proposals_idx = output_dict["proposals_idx"][:, 1].int().contiguous()
proposals_offset = output_dict["proposals_offset"]
# calculate iou of clustered instance
ious_on_cluster = common_ops.get_mask_iou_on_cluster(
proposals_idx, proposals_offset, data_dict["instance_ids"], data_dict["instance_num_point"]
)
# filter out background instances
fg_inds = (data_dict["instance_semantic_cls"] != -1)
fg_instance_cls = data_dict["instance_semantic_cls"][fg_inds]
fg_ious_on_cluster = ious_on_cluster[:, fg_inds]
# assign proposal to gt idx. -1: negative, 0 -> num_gts - 1: positive
num_proposals = fg_ious_on_cluster.size(0)
assigned_gt_inds = fg_ious_on_cluster.new_full((num_proposals,), -1, dtype=torch.long)
# overlap > thr on fg instances are positive samples
max_iou, argmax_iou = fg_ious_on_cluster.max(1)
pos_inds = max_iou >= self.hparams.cfg.model.network.train_cfg.pos_iou_thr
assigned_gt_inds[pos_inds] = argmax_iou[pos_inds]
"""classification loss"""
# follow detection convention: 0 -> K - 1 are fg, K is bg
labels = fg_instance_cls.new_full((num_proposals,), self.instance_classes)
pos_inds = assigned_gt_inds >= 0
labels[pos_inds] = fg_instance_cls[assigned_gt_inds[pos_inds]]
labels = labels.long()
losses["classification_loss"] = nn.functional.cross_entropy(output_dict["cls_scores"], labels)
"""mask scoring loss"""
mask_cls_label = labels[output_dict["instance_batch_idxs"].long()]
slice_inds = torch.arange(0, mask_cls_label.size(0), dtype=torch.long, device=mask_cls_label.device)
mask_scores_sigmoid_slice = output_dict["mask_scores"].sigmoid()[slice_inds, mask_cls_label]
mask_label, mask_label_mask = common_ops.get_mask_label(
proposals_idx, proposals_offset, data_dict["instance_ids"], data_dict["instance_semantic_cls"],
data_dict["instance_num_point"], ious_on_cluster,
-1, self.hparams.cfg.model.network.train_cfg.pos_iou_thr
)
mask_scoring_loss = nn.functional.binary_cross_entropy(
mask_scores_sigmoid_slice, mask_label.float(), weight=mask_label_mask, reduction="sum"
)
mask_scoring_loss /= (torch.count_nonzero(mask_label_mask) + 1)
losses["mask_scoring_loss"] = mask_scoring_loss
"""iou scoring loss"""
ious = common_ops.get_mask_iou_on_pred(
proposals_idx, proposals_offset, data_dict["instance_ids"], data_dict["instance_num_point"],
mask_scores_sigmoid_slice.detach()
)
slice_inds = torch.arange(0, labels.size(0), dtype=torch.long, device=labels.device)
iou_score_weight = labels < self.instance_classes
iou_score_slice = output_dict["iou_scores"][slice_inds, labels]
iou_scoring_loss = nn.functional.mse_loss(iou_score_slice, ious[:, fg_inds].max(1)[0], reduction="none")
losses["iou_scoring_loss"] = iou_scoring_loss[iou_score_weight].sum() / (iou_score_weight.count_nonzero() + 1)
return losses
def validation_step(self, data_dict, idx):
# prepare input and forward
output_dict = self(data_dict)
losses = self._loss(data_dict, output_dict)
# log losses
total_loss = 0
for loss_name, loss_value in losses.items():
total_loss += loss_value
self.log(f"val/{loss_name}", loss_value, on_step=False, on_epoch=True, sync_dist=True, batch_size=1)
self.log("val/total_loss", total_loss, on_step=False, on_epoch=True, sync_dist=True, batch_size=1)
# log semantic prediction accuracy
semantic_predictions = output_dict["semantic_scores"].max(1)[1]
semantic_accuracy = evaluate_semantic_accuracy(semantic_predictions, data_dict["sem_labels"], ignore_label=-1)
semantic_mean_iou = evaluate_semantic_miou(semantic_predictions, data_dict["sem_labels"], ignore_label=-1)
self.log(
"val_eval/semantic_accuracy", semantic_accuracy, on_step=False, on_epoch=True, sync_dist=True, batch_size=1
)
self.log(
"val_eval/semantic_mean_iou", semantic_mean_iou, on_step=False, on_epoch=True, sync_dist=True, batch_size=1
)
if self.current_epoch > self.hparams.cfg.model.network.prepare_epochs:
point_xyz_cpu = data_dict["point_xyz"].cpu().numpy()
instance_ids_cpu = data_dict["instance_ids"].cpu()
sem_labels = data_dict["sem_labels"].cpu()
pred_instances = self._get_pred_instances(
data_dict["scan_ids"][0], point_xyz_cpu, output_dict["proposals_idx"].cpu(),
output_dict["semantic_scores"].size(0), output_dict["cls_scores"].cpu(),
output_dict["iou_scores"].cpu(), output_dict["mask_scores"].cpu(),
len(self.hparams.cfg.data.ignore_classes)
)
gt_instances = get_gt_instances(
sem_labels, instance_ids_cpu, self.hparams.cfg.data.ignore_classes
)
gt_instances_bbox = get_gt_bbox(
point_xyz_cpu, instance_ids_cpu.numpy(),
sem_labels.numpy(), -1, self.hparams.cfg.data.ignore_classes
)
self.val_test_step_outputs.append((pred_instances, gt_instances, gt_instances_bbox))
def test_step(self, data_dict, idx):
# prepare input and forward
output_dict = self(data_dict)
semantic_accuracy = None
semantic_mean_iou = None
if self.hparams.cfg.model.inference.evaluate:
semantic_predictions = output_dict["semantic_scores"].max(1)[1]
semantic_accuracy = evaluate_semantic_accuracy(
semantic_predictions, data_dict["sem_labels"], ignore_label=-1
)
semantic_mean_iou = evaluate_semantic_miou(
semantic_predictions, data_dict["sem_labels"], ignore_label=-1
)
if self.current_epoch > self.hparams.cfg.model.network.prepare_epochs:
point_xyz_cpu = data_dict["point_xyz"].cpu().numpy()
instance_ids_cpu = data_dict["instance_ids"].cpu()
sem_labels = data_dict["sem_labels"].cpu()
pred_instances = self._get_pred_instances(
data_dict["scan_ids"][0], point_xyz_cpu, output_dict["proposals_idx"].cpu(),
output_dict["semantic_scores"].size(0), output_dict["cls_scores"].cpu(), output_dict["iou_scores"].cpu(),
output_dict["mask_scores"].cpu(), len(self.hparams.cfg.data.ignore_classes)
)
gt_instances = None
gt_instances_bbox = None
if self.hparams.cfg.model.inference.evaluate:
gt_instances = get_gt_instances(
sem_labels, instance_ids_cpu, self.hparams.cfg.data.ignore_classes
)
gt_instances_bbox = get_gt_bbox(
point_xyz_cpu, instance_ids_cpu.numpy(),
sem_labels.numpy(), -1, self.hparams.cfg.data.ignore_classes
)
self.val_test_step_outputs.append(
(semantic_accuracy, semantic_mean_iou, pred_instances, gt_instances, gt_instances_bbox)
)
def _get_pred_instances(self, scan_id, gt_xyz, proposals_idx, num_points, cls_scores, iou_scores, mask_scores,
num_ignored_classes):
num_instances = cls_scores.size(0)
cls_scores = cls_scores.softmax(1)
cls_pred_list, score_pred_list, mask_pred_list = [], [], []
for i in range(self.instance_classes):
cls_pred = cls_scores.new_full((num_instances,), i + 1, dtype=torch.long)
cur_cls_scores = cls_scores[:, i]
cur_iou_scores = iou_scores[:, i]
cur_mask_scores = mask_scores[:, i]
score_pred = cur_cls_scores * cur_iou_scores.clamp(0, 1)
mask_pred = torch.zeros((num_instances, num_points), dtype=torch.bool, device="cpu")
mask_inds = cur_mask_scores > self.hparams.cfg.model.network.test_cfg.mask_score_thr
cur_proposals_idx = proposals_idx[mask_inds]
mask_pred[cur_proposals_idx[:, 0], cur_proposals_idx[:, 1]] = True
# filter low score instance
inds = cur_cls_scores > self.hparams.cfg.model.network.test_cfg.cls_score_thr
cls_pred = cls_pred[inds]
score_pred = score_pred[inds]
mask_pred = mask_pred[inds]
# filter too small instances
npoint = torch.count_nonzero(mask_pred, dim=1)
inds = npoint >= self.hparams.cfg.model.network.test_cfg.min_npoint
cls_pred = cls_pred[inds]
score_pred = score_pred[inds]
mask_pred = mask_pred[inds]
cls_pred_list.append(cls_pred)
score_pred_list.append(score_pred)
mask_pred_list.append(mask_pred)
cls_pred = torch.cat(cls_pred_list).numpy()
score_pred = torch.cat(score_pred_list).numpy()
mask_pred = torch.cat(mask_pred_list).numpy()
pred_instances = []
for i in range(cls_pred.shape[0]):
pred = {'scan_id': scan_id, 'label_id': cls_pred[i], 'conf': score_pred[i],
'pred_mask': rle_encode(mask_pred[i])}
pred_xyz = gt_xyz[mask_pred[i]]
pred['pred_bbox'] = np.concatenate((pred_xyz.min(0), pred_xyz.max(0)))
pred_instances.append(pred)
return pred_instances