-
Notifications
You must be signed in to change notification settings - Fork 12
/
evaluate.py
510 lines (451 loc) · 23.2 KB
/
evaluate.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
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
# Copyright (c) 2022 IDEA. All Rights Reserved.
# ------------------------------------------------------------------------
import argparse
import datetime
import json
import random
import time
from pathlib import Path
import os, sys
import numpy as np
from tqdm import tqdm
import torch
from collections import defaultdict
from typing import Iterable, Mapping, Tuple, Union
from util.slconfig import DictAction, SLConfig
import util.misc as utils
from datasets import build_dataset, get_coco_api_from_dataset
def dict2string(dict):
string = ""
for k, v in dict.items():
string += f"{k}: {v} \n"
return string
def get_args_parser():
parser = argparse.ArgumentParser('Set transformer detector', add_help=False)
parser.add_argument('--config_file', '-c', type=str, required=True)
parser.add_argument('--options',
nargs='+',
action=DictAction,
help='override some settings in the used config, the key-value pair '
'in xxx=yyy format will be merged into config file.')
# dataset parameters
parser.add_argument('--dataset_file', default='coco')
parser.add_argument('--eval_checkpoint', type=str, default='/comp_robot/cv_public_dataset/COCO2017/')
parser.add_argument('--coco_path', type=str, default='/comp_robot/cv_public_dataset/COCO2017/')
parser.add_argument('--data_path', type=str, default='/path/to/a/specific/datasets')
parser.add_argument('--coco_panoptic_path', type=str)
parser.add_argument('--remove_difficult', action='store_true')
parser.add_argument('--fix_size', action='store_true')
# training parameters
parser.add_argument('--output_dir', default='',
help='path where to save, empty for no saving')
parser.add_argument('--note', default='',
help='add some notes to the experiment')
parser.add_argument('--device', default='cuda',
help='device to use for training / testing')
parser.add_argument('--seed', default=42, type=int)
parser.add_argument('--resume', default='', help='resume from checkpoint')
parser.add_argument('--pretrain_model_path', help='load from other checkpoint')
parser.add_argument('--finetune_ignore', type=str, nargs='+')
parser.add_argument('--start_epoch', default=0, type=int, metavar='N',
help='start epoch')
parser.add_argument('--eval', action='store_true')
parser.add_argument('--num_workers', default=10, type=int)
parser.add_argument('--test', action='store_true')
parser.add_argument('--find_unused_params', action='store_true')
parser.add_argument('--save_results', action='store_true')
parser.add_argument('--save_log', action='store_true')
# distributed training parameters
parser.add_argument('--world_size', default=1, type=int,
help='number of distributed processes')
parser.add_argument('--dist_url', default='env://', help='url used to set up distributed training')
parser.add_argument('--rank', default=0, type=int,
help='number of distributed processes')
parser.add_argument("--local_rank", type=int, help='local rank for DistributedDataParallel')
parser.add_argument('--amp', action='store_true',
help="Train with mixed precision")
return parser
def compute_tapvid_metrics(
query_frame: np.ndarray,
gt_occluded: np.ndarray,
gt_tracks: np.ndarray,
pred_occluded: np.ndarray,
pred_tracks: np.ndarray,
query_mode: str,
) -> Mapping[str, np.ndarray]:
"""Computes TAP-Vid metrics (Jaccard, Pts. Within Thresh, Occ. Acc.)
See the TAP-Vid paper for details on the metric computation. All inputs are
given in raster coordinates. The first three arguments should be the direct
outputs of the reader: the 'query_points', 'occluded', and 'target_points'.
The paper metrics assume these are scaled relative to 256x256 images.
pred_occluded and pred_tracks are your algorithm's predictions.
This function takes a batch of inputs, and computes metrics separately for
each video. The metrics for the full benchmark are a simple mean of the
metrics across the full set of videos. These numbers are between 0 and 1,
but the paper multiplies them by 100 to ease reading.
Args:
query_frame: The start tracking frame. Its size is
[b, n], where b is the batch size and n is the number of queries
gt_occluded: A boolean array of shape [b, n, t], where t is the number
of frames. True indicates that the point is occluded.
gt_tracks: The target points, of shape [b, n, t, 2]. Each point is
in the format [x, y]
pred_occluded: A boolean array of predicted occlusions, in the same
format as gt_occluded.
pred_tracks: An array of track predictions from your algorithm, in the
same format as gt_tracks.
query_mode: Either 'first' or 'strided', depending on how queries are
sampled. If 'first', we assume the prior knowledge that all points
before the query point are occluded, and these are removed from the
evaluation.
Returns:
A dict with the following keys:
occlusion_accuracy: Accuracy at predicting occlusion.
pts_within_{x} for x in [1, 2, 4, 8, 16]: Fraction of points
predicted to be within the given pixel threshold, ignoring occlusion
prediction.
jaccard_{x} for x in [1, 2, 4, 8, 16]: Jaccard metric for the given
threshold
average_pts_within_thresh: average across pts_within_{x}
average_jaccard: average across jaccard_{x}
"""
metrics = {}
# Fixed bug is described in:
# https://github.com/facebookresearch/co-tracker/issues/20
eye = np.eye(gt_tracks.shape[2], dtype=np.int32)
if query_mode == "first":
# evaluate frames after the query frame
query_frame_to_eval_frames = np.cumsum(eye, axis=1) - eye
elif query_mode == "strided":
# evaluate all frames except the query frame
query_frame_to_eval_frames = 1 - eye
else:
raise ValueError("Unknown query mode " + query_mode)
query_frame = np.round(query_frame).astype(np.int32)
evaluation_points = query_frame_to_eval_frames[query_frame] > 0
# Occlusion accuracy is simply how often the predicted occlusion equals the
# ground truth.
occ_acc = np.sum(
np.equal(pred_occluded, gt_occluded) & evaluation_points,
axis=(1, 2),
) / np.sum(evaluation_points)
metrics["occlusion_accuracy"] = occ_acc
# Next, convert the predictions and ground truth positions into pixel
# coordinates.
visible = np.logical_not(gt_occluded)
pred_visible = np.logical_not(pred_occluded)
all_frac_within = []
all_jaccard = []
for thresh in [1, 2, 4, 8, 16]:
# True positives are points that are within the threshold and where both
# the prediction and the ground truth are listed as visible.
within_dist = np.sum(
np.square(pred_tracks - gt_tracks),
axis=-1,
) < np.square(thresh)
is_correct = np.logical_and(within_dist, visible)
# Compute the frac_within_threshold, which is the fraction of points
# within the threshold among points that are visible in the ground truth,
# ignoring whether they're predicted to be visible.
count_correct = np.sum(
is_correct & evaluation_points,
axis=(1, 2),
)
count_visible_points = np.sum(visible & evaluation_points, axis=(1, 2))
frac_correct = count_correct / count_visible_points
metrics["pts_within_" + str(thresh)] = frac_correct
all_frac_within.append(frac_correct)
true_positives = np.sum(
is_correct & pred_visible & evaluation_points, axis=(1, 2)
)
# The denominator of the jaccard metric is the true positives plus
# false positives plus false negatives. However, note that true positives
# plus false negatives is simply the number of points in the ground truth
# which is easier to compute than trying to compute all three quantities.
# Thus we just add the number of points in the ground truth to the number
# of false positives.
#
# False positives are simply points that are predicted to be visible,
# but the ground truth is not visible or too far from the prediction.
gt_positives = np.sum(visible & evaluation_points, axis=(1, 2))
false_positives = (~visible) & pred_visible
false_positives = false_positives | ((~within_dist) & pred_visible)
false_positives = np.sum(false_positives & evaluation_points, axis=(1, 2))
jaccard = true_positives / (gt_positives + false_positives)
metrics["jaccard_" + str(thresh)] = jaccard
all_jaccard.append(jaccard)
metrics["average_jaccard"] = np.mean(
np.stack(all_jaccard, axis=1),
axis=1,
)
metrics["average_pts_within_thresh"] = np.mean(
np.stack(all_frac_within, axis=1),
axis=1,
)
return metrics
def reduce_masked_mean(input, mask, dim=None, keepdim=False):
r"""Masked mean
`reduce_masked_mean(x, mask)` computes the mean of a tensor :attr:`input`
over a mask :attr:`mask`, returning
.. math::
\text{output} =
\frac
{\sum_{i=1}^N \text{input}_i \cdot \text{mask}_i}
{\epsilon + \sum_{i=1}^N \text{mask}_i}
where :math:`N` is the number of elements in :attr:`input` and
:attr:`mask`, and :math:`\epsilon` is a small constant to avoid
division by zero.
`reduced_masked_mean(x, mask, dim)` computes the mean of a tensor
:attr:`input` over a mask :attr:`mask` along a dimension :attr:`dim`.
Optionally, the dimension can be kept in the output by setting
:attr:`keepdim` to `True`. Tensor :attr:`mask` must be broadcastable to
the same dimension as :attr:`input`.
The interface is similar to `torch.mean()`.
Args:
inout (Tensor): input tensor.
mask (Tensor): mask.
dim (int, optional): Dimension to sum over. Defaults to None.
keepdim (bool, optional): Keep the summed dimension. Defaults to False.
Returns:
Tensor: mean tensor.
"""
mask = mask.expand_as(input)
prod = input * mask
if dim is None:
numer = torch.sum(prod)
denom = torch.sum(mask)
else:
numer = torch.sum(prod, dim=dim, keepdim=keepdim)
denom = torch.sum(mask, dim=dim, keepdim=keepdim)
EPS = 1e-6
mean = numer / (EPS + denom)
return mean
class Evaluator:
"""
A class defining evaluator, refer from CoTrackerv2.
"""
def __init__(self, exp_dir, ckpt_name) -> None:
# Visualization
self.exp_dir = exp_dir
os.makedirs(exp_dir, exist_ok=True)
self.visualization_filepaths = defaultdict(lambda: defaultdict(list))
self.visualize_dir = os.path.join(exp_dir, "visualisations", ckpt_name)
if not os.path.exists(self.visualize_dir):
os.makedirs(self.visualize_dir)
self.test_log_dir = os.path.join(exp_dir, "test", ckpt_name)
if not os.path.exists(self.test_log_dir):
os.makedirs(self.test_log_dir)
self.results_AJ_ours = {}
self.results_AJ_comp = {}
self.results_DeltaX_ours = {}
self.results_DeltaX_comp = {}
def compute_metrics(self, metrics, seq_name, targets, predictions, dataset_name, res_H, res_W):
"""compute metrics for different datasets according to the predictions and targets.
Args:
metrics (dict): recording the results of different metrics for each sequence.
seq_name (str): the name of the currently evaluating sequence.
targets (dict):
pt_boxes (torch.FloatTensor): [num_queries, num_frames, 4], the gt trajectory of queries.
pt_labels (torch.FloatTensor): [num_queries, num_frames], the gt visibility of queries (1/0).
query_frames (torch.IntTensor): [num_queries], the first emerge frame.
predictions (dict):
pred_boxes (torch.FloatTensor): [num_queries, num_frames, 4], the predicted trajectory of queries.
pred_labels (torch.FloatTensor): [num_queries, num_frames], the predicted visibility of queries (1/0).
dataset_name (_type_): _description_
res_H, res_W (Int): the resolution of the video.
"""
num_queries = targets["num_real_pt"]
pred_trajectory, pred_visibility = predictions["pred_boxes"][:num_queries], predictions["pred_labels"][:num_queries]
pred_tracks = (pred_trajectory[None, ..., :2] * pred_trajectory.new_tensor([res_W-1, res_H-1])).cpu().numpy()
pred_occluded = (pred_visibility != 1)[None, ...].cpu().numpy()
gt_trajectory, gt_visibility, query_frames = targets["pt_boxes"][:num_queries], targets["pt_labels"][:num_queries], targets["query_frames"][:num_queries]
gt_tracks = (gt_trajectory[None, ..., :2] * pred_trajectory.new_tensor([res_W-1, res_H-1])).cpu().numpy()
gt_occluded = (gt_visibility != 1)[None, ...].cpu().numpy()
query_frames = query_frames[None, :].cpu().numpy()
if "tapvid" in dataset_name:
out_metrics = compute_tapvid_metrics(
query_frames,
gt_occluded,
gt_tracks,
pred_occluded,
pred_tracks,
query_mode="strided" if "strided" in dataset_name else "first",
)
metrics[seq_name] = out_metrics
for metric_name in out_metrics.keys():
out_metrics[metric_name] = out_metrics[metric_name][0]
if "avg" not in metrics:
metrics["avg"] = {}
metrics["avg"][metric_name] = np.mean(
[v[metric_name] for k, v in metrics.items() if k != "avg"]
)
print("Sequence Name: ", seq_name)
print(f"Metrics: {out_metrics}")
print(f"\nAvg Metric: {metrics['avg']}")
else:
raise NotImplementedError
def build_model_main(args):
# we use register to maintain models from catdet6 on.
from models.registry import MODULE_BUILD_FUNCS
assert args.modelname in MODULE_BUILD_FUNCS._module_dict
build_func = MODULE_BUILD_FUNCS.get(args.modelname)
model, criterion, postprocessors = build_func(args)
return model, criterion, postprocessors
def summarize_all_points(outputs, targets):
outputs_sum = {}
targets_sum = {}
# full_seq_output
outputs_sum["full_seq_output"] = {
"pred_logits": [],
"pred_boxes": [],
"dn_meta": None
}
targets_sum["full_seq_target"] = [{
"pt_boxes": [],
"pt_labels": [],
"pt_tracking_mask": [],
"ptq_update_mask": [],
"query_frames": [],
"num_real_pt": 0,
}]
for output, target in zip(outputs, targets):
num_real_pt = target["full_seq_target"][0]["num_real_pt"]
outputs_sum["full_seq_output"]["pred_logits"].append(output["full_seq_output"]["pred_logits"][:, :, :num_real_pt])
outputs_sum["full_seq_output"]["pred_boxes"].append(output["full_seq_output"]["pred_boxes"][:, :, :num_real_pt])
targets_sum["full_seq_target"][0]["pt_boxes"].append(target["full_seq_target"][0]["pt_boxes"][:num_real_pt])
targets_sum["full_seq_target"][0]["pt_labels"].append(target["full_seq_target"][0]["pt_labels"][:num_real_pt])
targets_sum["full_seq_target"][0]["pt_tracking_mask"].append(target["full_seq_target"][0]["pt_tracking_mask"][:num_real_pt])
targets_sum["full_seq_target"][0]["ptq_update_mask"].append(target["full_seq_target"][0]["ptq_update_mask"][:num_real_pt])
targets_sum["full_seq_target"][0]["query_frames"].append(target["full_seq_target"][0]["query_frames"][:num_real_pt])
targets_sum["full_seq_target"][0]["num_real_pt"] += num_real_pt
outputs_sum["full_seq_output"]["pred_logits"] = torch.cat(outputs_sum["full_seq_output"]["pred_logits"], dim=2)
outputs_sum["full_seq_output"]["pred_boxes"] = torch.cat(outputs_sum["full_seq_output"]["pred_boxes"], dim=2)
targets_sum["full_seq_target"][0]["pt_boxes"] = torch.cat(targets_sum["full_seq_target"][0]["pt_boxes"], dim=0)
targets_sum["full_seq_target"][0]["pt_labels"] = torch.cat(targets_sum["full_seq_target"][0]["pt_labels"], dim=0)
targets_sum["full_seq_target"][0]["pt_tracking_mask"] = torch.cat(targets_sum["full_seq_target"][0]["pt_tracking_mask"], dim=0)
targets_sum["full_seq_target"][0]["ptq_update_mask"] = torch.cat(targets_sum["full_seq_target"][0]["ptq_update_mask"], dim=0)
targets_sum["full_seq_target"][0]["query_frames"] = torch.cat(targets_sum["full_seq_target"][0]["query_frames"], dim=0)
return outputs_sum, targets_sum
def main(args):
# utils.init_distributed_mode(args)
# load cfg file and update the args
print("Loading config file from {}".format(args.config_file))
time.sleep(args.rank * 0.02)
cfg = SLConfig.fromfile(args.config_file)
if args.options is not None:
cfg.merge_from_dict(args.options)
cfg_dict = cfg._cfg_dict.to_dict()
args_vars = vars(args)
for k,v in cfg_dict.items():
if k not in args_vars:
setattr(args, k, v)
else:
raise ValueError("Key {} can used by args only".format(k))
print(args)
device = torch.device(args.device)
# fix the seed for reproducibility
seed = args.seed + utils.get_rank()
torch.manual_seed(seed)
np.random.seed(seed)
random.seed(seed)
# build model
model, _, _ = build_model_main(args)
if not args.eval_checkpoint:
args.eval_checkpoint = os.path.join(args.output_dir, 'checkpoint.pth')
checkpoint = torch.load(args.eval_checkpoint, map_location='cpu')
epoch = checkpoint['epoch']
model.to(device)
model.eval()
print("loading checkpoint from {}".format(args.eval_checkpoint))
use_ema_model = getattr(args, 'use_ema', False)
if not use_ema_model:
model.load_state_dict(checkpoint['model'])
else:
model_state_dict = {}
for name, value in checkpoint["ema_model"].items():
model_state_dict[name.replace("module.", "")] = value
model.load_state_dict(model_state_dict)
# dataset
dataset_val = build_dataset(image_set=args.dataset_file, args=args) # val
# Evaluator
evaluator = Evaluator(args.output_dir, args.eval_checkpoint.split("/")[-1].split(".")[0])
# Run evaluate.
metrics = {}
for data_id in tqdm(range(len(dataset_val))):
if not ("strided" in args.dataset_file): # first.
(samples, targets, seq_name) = dataset_val[data_id]
H, W = 256, 256
if isinstance(samples, (list, torch.Tensor)):
from util.misc import nested_temporal_tensor_from_tensor_list
samples = nested_temporal_tensor_from_tensor_list(samples[None, ...])
samples = samples.to(device)
targets = {k: v.to(device) for k, v in targets.items()}
with torch.no_grad():
outputs, targets = model(samples, [targets])
else: # strided
def fuse_bidir_outputs(outputs_, outputs_inv_, targets_):
# fuse the bidirectional outputs, to mimic the offline tracker's results.
outputs_forward_backward = {
"full_seq_output": {
"dn_meta": None,
}
}
mask_forward = targets_["full_seq_target"][0]["pt_tracking_mask"].float().transpose(0,1) # T N
mask_backward = 1 - mask_forward
outputs_forward_backward["full_seq_output"]["pred_logits"] = \
(outputs_["full_seq_output"]["pred_logits"] * mask_forward[None, ..., None]) + (outputs_inv_["full_seq_output"]["pred_logits"].flip(1) * mask_backward[None, ..., None])
outputs_forward_backward["full_seq_output"]["pred_boxes"] = \
(outputs_["full_seq_output"]["pred_boxes"] * mask_forward[None, ..., None]) + (outputs_inv_["full_seq_output"]["pred_boxes"].flip(1) * mask_backward[None, ..., None])
return outputs_forward_backward
strided_data_list, strided_inv_data_list = dataset_val[data_id]
targets_bi = []
outputs_bi = []
for ((samples, targets_, seq_name_), (samples_inv, targets_inv_, seq_name_inv_)) in zip(strided_data_list, strided_inv_data_list):
H, W = 256, 256
if isinstance(samples, (list, torch.Tensor)):
from util.misc import nested_temporal_tensor_from_tensor_list
samples = nested_temporal_tensor_from_tensor_list(samples[None, ...])
samples_inv = nested_temporal_tensor_from_tensor_list(samples_inv[None, ...])
samples = samples.to(device)
samples_inv = samples_inv.to(device)
targets_ = {k: v.to(device) for k, v in targets_.items()}
targets_inv_ = {k: v.to(device) for k, v in targets_inv_.items()}
with torch.no_grad():
print("processing: ", seq_name_)
outputs_, targets_ = model(samples, [targets_])
print("processing: ", seq_name_inv_)
outputs_inv_, targets_inv_ = model(samples_inv, [targets_inv_])
outputs_bi_ = fuse_bidir_outputs(outputs_, outputs_inv_, targets_)
targets_bi_ = targets_
# targets_bi_["full_seq_target"][0]["query_frames"] *= 0 # offline tracker. but still need to know which frame is the reference frame.
outputs_bi.append(outputs_bi_)
targets_bi.append(targets_bi_)
seq_name = seq_name_
outputs, targets = summarize_all_points(outputs_bi, targets_bi)
outputs["window_output_list"] = []
targets["window_target_list"] = []
outputs = outputs["full_seq_output"]
outputs["pred_boxes"] = outputs["pred_boxes"][0].permute(1, 0, 2)
threshold_occ = 0.5
pred_occluded = outputs["pred_logits"][0].permute(1, 0, 2)[..., 1].sigmoid() > threshold_occ
outputs["pred_labels"] = pred_occluded
evaluator.compute_metrics(metrics, seq_name, targets["full_seq_target"][0], outputs, args.dataset_file, H, W)
# Saving the evaluation results to a .log file
evaluate_result = metrics.pop("avg")
result_file = os.path.join(args.output_dir, f"test/{args.eval_checkpoint.split('/')[-1].split('.')[0]}.log")
print(f"Dumping eval results to {result_file}.")
with open(result_file, "w") as f:
for name, value in metrics.items():
f.write(f"{name}:\n {dict2string(value)} \n")
all_result = "\n========= All Results \n"
all_result += dict2string(evaluate_result)
main_result = "\n========= Main Results \n" + f"AJ : {evaluate_result['average_jaccard']} \nDelta_x: {evaluate_result['average_pts_within_thresh']} \nOA : {evaluate_result['occlusion_accuracy']}"
print(main_result)
f.write(all_result)
f.write(main_result)
if __name__ == '__main__':
parser = argparse.ArgumentParser('TAPTR evaluation script', parents=[get_args_parser()])
args = parser.parse_args()
if args.output_dir:
Path(args.output_dir).mkdir(parents=True, exist_ok=True)
main(args)