File size: 6,638 Bytes
3bbb319 | 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 | # Copyright (c) OpenMMLab. All rights reserved.
import argparse
import copy
import os
import time
import torch
from mmcv import Config, DictAction
from mmcv.cnn import fuse_conv_bn
from mmcv.parallel import MMDistributedDataParallel
from mmcv.runner import init_dist, load_checkpoint, wrap_fp16_model
from mmdet.datasets import (build_dataloader, build_dataset,
replace_ImageToTensor)
from mmdet.models import build_detector
from mmdet.utils import replace_cfg_vals, update_data_root
def parse_args():
parser = argparse.ArgumentParser(description='MMDet benchmark a model')
parser.add_argument('config', help='test config file path')
parser.add_argument('checkpoint', help='checkpoint file')
parser.add_argument(
'--repeat-num',
type=int,
default=1,
help='number of repeat times of measurement for averaging the results')
parser.add_argument(
'--max-iter', type=int, default=2000, help='num of max iter')
parser.add_argument(
'--log-interval', type=int, default=50, help='interval of logging')
parser.add_argument(
'--fuse-conv-bn',
action='store_true',
help='Whether to fuse conv and bn, this will slightly increase'
'the inference speed')
parser.add_argument(
'--cfg-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. If the value to '
'be overwritten is a list, it should be like key="[a,b]" or key=a,b '
'It also allows nested list/tuple values, e.g. key="[(a,b),(c,d)]" '
'Note that the quotation marks are necessary and that no white space '
'is allowed.')
parser.add_argument(
'--launcher',
choices=['none', 'pytorch', 'slurm', 'mpi'],
default='none',
help='job launcher')
parser.add_argument('--local_rank', type=int, default=0)
args = parser.parse_args()
if 'LOCAL_RANK' not in os.environ:
os.environ['LOCAL_RANK'] = str(args.local_rank)
return args
def measure_inference_speed(cfg, checkpoint, max_iter, log_interval,
is_fuse_conv_bn):
# set cudnn_benchmark
if cfg.get('cudnn_benchmark', False):
torch.backends.cudnn.benchmark = True
cfg.model.pretrained = None
cfg.data.test.test_mode = True
# build the dataloader
samples_per_gpu = cfg.data.test.pop('samples_per_gpu', 1)
if samples_per_gpu > 1:
# Replace 'ImageToTensor' to 'DefaultFormatBundle'
cfg.data.test.pipeline = replace_ImageToTensor(cfg.data.test.pipeline)
dataset = build_dataset(cfg.data.test)
data_loader = build_dataloader(
dataset,
samples_per_gpu=1,
# Because multiple processes will occupy additional CPU resources,
# FPS statistics will be more unstable when workers_per_gpu is not 0.
# It is reasonable to set workers_per_gpu to 0.
workers_per_gpu=0,
dist=True,
shuffle=False)
# build the model and load checkpoint
cfg.model.train_cfg = None
model = build_detector(cfg.model, test_cfg=cfg.get('test_cfg'))
fp16_cfg = cfg.get('fp16', None)
if fp16_cfg is not None:
wrap_fp16_model(model)
load_checkpoint(model, checkpoint, map_location='cpu')
if is_fuse_conv_bn:
model = fuse_conv_bn(model)
model = MMDistributedDataParallel(
model.cuda(),
device_ids=[torch.cuda.current_device()],
broadcast_buffers=False)
model.eval()
# the first several iterations may be very slow so skip them
num_warmup = 5
pure_inf_time = 0
fps = 0
# benchmark with 2000 image and take the average
for i, data in enumerate(data_loader):
torch.cuda.synchronize()
start_time = time.perf_counter()
with torch.no_grad():
model(return_loss=False, rescale=True, **data)
torch.cuda.synchronize()
elapsed = time.perf_counter() - start_time
if i >= num_warmup:
pure_inf_time += elapsed
if (i + 1) % log_interval == 0:
fps = (i + 1 - num_warmup) / pure_inf_time
print(
f'Done image [{i + 1:<3}/ {max_iter}], '
f'fps: {fps:.1f} img / s, '
f'times per image: {1000 / fps:.1f} ms / img',
flush=True)
if (i + 1) == max_iter:
fps = (i + 1 - num_warmup) / pure_inf_time
print(
f'Overall fps: {fps:.1f} img / s, '
f'times per image: {1000 / fps:.1f} ms / img',
flush=True)
break
return fps
def repeat_measure_inference_speed(cfg,
checkpoint,
max_iter,
log_interval,
is_fuse_conv_bn,
repeat_num=1):
assert repeat_num >= 1
fps_list = []
for _ in range(repeat_num):
#
cp_cfg = copy.deepcopy(cfg)
fps_list.append(
measure_inference_speed(cp_cfg, checkpoint, max_iter, log_interval,
is_fuse_conv_bn))
if repeat_num > 1:
fps_list_ = [round(fps, 1) for fps in fps_list]
times_pre_image_list_ = [round(1000 / fps, 1) for fps in fps_list]
mean_fps_ = sum(fps_list_) / len(fps_list_)
mean_times_pre_image_ = sum(times_pre_image_list_) / len(
times_pre_image_list_)
print(
f'Overall fps: {fps_list_}[{mean_fps_:.1f}] img / s, '
f'times per image: '
f'{times_pre_image_list_}[{mean_times_pre_image_:.1f}] ms / img',
flush=True)
return fps_list
return fps_list[0]
def main():
args = parse_args()
cfg = Config.fromfile(args.config)
# replace the ${key} with the value of cfg.key
cfg = replace_cfg_vals(cfg)
# update data root according to MMDET_DATASETS
update_data_root(cfg)
if args.cfg_options is not None:
cfg.merge_from_dict(args.cfg_options)
if args.launcher == 'none':
raise NotImplementedError('Only supports distributed mode')
else:
init_dist(args.launcher, **cfg.dist_params)
repeat_measure_inference_speed(cfg, args.checkpoint, args.max_iter,
args.log_interval, args.fuse_conv_bn,
args.repeat_num)
if __name__ == '__main__':
main()
|