1.环境配置
| 操作 | 命令 | 版本 |
|---|---|---|
| 检查cuda版本 | nvcc -V | 11.4 |
| 根据cuda版本下载pytorch（官网：https://pytorch.org/get-started/locally/） | pip3 install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113 | 1.10.2 |
| 根据pytorch版本下载mmcv | pip install mmcv-full==1.4.6 -f https://download.openmmlab.com/mmcv/dist/cu113/torch1.10.0/index.html | 1.4.6 |
| 下载mmdetection(mmcv与mmdetection版本要对应：https://github.com/open-mmlab/mmdetection/blob/master/docs/zh_cn/get_started.md) | git clone https://github.com/open-mmlab/mmdetection.git cd mmdetection pip install -r requirements/build.txt pip install -e . | 2.23.0 |
2.测试环境是否安装正确
import os
import sys
from mmdet.apis import init_detector,inference_detector, show_result_pyplot
config_file = './configs/faster_rcnn/faster_rcnn_r50_fpn_1x_coco.py'
checkpoint_file = 'faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth'
device='cuda:0'
model = init_detector(config_file,checkpoint_file,device=device)
img = './demo/demo.jpg'
result = inference_detector(model, img)
show_result_pyplot(model, img, result, score_thr=0.3)
3.打印出你需要的config文件
- 查看config文件：
python ./tools/misc/print_config.py ./configs/yolox/yolox_l_8x8_300e_coco.py>>yolox_l_8x8_300e_coco.txt
并将其修改为自己的config文件：
- 删除首行“Config:”
- load_from：预训练模型路径
- resume_from：紧接上一次训练的模型路径
- num_classes：类别个数
- ann_file：json文件路径
- img_prefix：jpg文件路径
- persistent_workers=False 不开多进程
- max_epochs=300
- samples_per_gpu：batchsize的个数
- workers_per_gpu：多进程的个数，可设为0
- work_dir：模型权重及log的保存位置
-
修改/mmdet/datasets/coco.py：
CLASSES = (“person”,)
PALETTE = (220, 20, 60) -
修改/mmdet/core/evaluation/class_names.py：
return [‘person’] -
修改/tools/train.py
validate = True
4.开启多进程
要想给num_workers设置较大的值，必须在创建docker时配置较大的共享内存，加入参数–shm-size=“15g”，这时workers_per_gpu=10，persistent_workers=True。（如有迷惑请看其他博文）
5.开始训练
单gpu训练：
python ./tools/train.py ./configs/yolox/my_yolox_l.py --gpu-id 2
多gpu训练（未成功）：
bash ./tools/dist_train.sh ./configs/yolox/my_yolox_l.py 2
6.绘制曲线
| 评价指标 | 命令 |
|---|---|
| map | python ./tools/analysis_tools/analyze_logs.py plot_curve ./tutorial_exps/yolox_l_8x8_300e_coco/backup_2/20220511_030327.log.json --keys mAP --legend mAP_bbox |
| loss | python ./tools/analysis_tools/analyze_logs.py plot_curve ./tutorial_exps/yolof_r50_c5_8x8_1x_coco/backup_2/20220516_023437.log.json --keys loss_cls loss_bbox --legend loss_cls loss_bbox --start-epoch 2 --out losses.jpg |
此处略微修改了一下analyze_logs.py文件，其中注释掉的内容为修改处，详见下面代码，主要目的是让–start-epoch有效。
def plot_curve(log_dicts, args):
if args.backend is not None:
plt.switch_backend(args.backend)
sns.set_style(args.style)
# if legend is None, use {filename}_{key} as legend
legend = args.legend
if legend is None:
legend = []
for json_log in args.json_logs:
for metric in args.keys:
legend.append(f'{
json_log}_{
metric}')
assert len(legend) == (len(args.json_logs) * len(args.keys))
metrics = args.keys
print("**************",metrics)
num_metrics = len(metrics)
for i, log_dict in enumerate(log_dicts):
epochs = list(log_dict.keys())
for j, metric in enumerate(metrics):
print(f'plot curve of {
args.json_logs[i]}, metric is {
metric}')
if metric not in log_dict[epochs[int(args.start_epoch) - 1]]:
if 'bbox_mAP' in metric:
raise KeyError(
f'{
args.json_logs[i]} does not contain metric '
f'{
metric}. Please check if "--no-validate" is '
'specified when you trained the model.')
raise KeyError(
f'{
args.json_logs[i]} does not contain metric {
metric}. '
'Please reduce the log interval in the config so that '
'interval is less than iterations of one epoch.')
if 'bbox_mAP' in metric:
# xs = np.arange(
# int(args.start_epoch),
# max(epochs), int(args.eval_interval))
xs = []
ys = []
# for epoch in epochs:
# ys += log_dict[epoch][metric]
print(args.eval_interval)
print("epochs=",epochs)
# print(xs,range(int(args.start_epoch),max(epochs)+1, int(args.eval_interval)))
for x in range(int(args.start_epoch),max(epochs), int(args.eval_interval)):
if log_dict[x][metric]:
ys += log_dict[x][metric]
xs.append(x)
ax = plt.gca()
ax.set_xticks(xs)
plt.xlabel('epoch')
plt.plot(xs, ys,label=legend[i * num_metrics + j], marker='o')
else:
xs = []
ys = []
num_iters_per_epoch = log_dict[int(args.start_epoch)]['iter'][-2]#log_dict[epochs[0]]['iter'][-2]
for x in range(int(args.start_epoch),max(epochs), int(args.eval_interval)):
iters = log_dict[x]['iter']
if log_dict[x]['mode'][-1] == 'val':
iters = iters[:-1]
xs.append(
np.array(iters) + (x - 1) * num_iters_per_epoch)
ys.append(np.array(log_dict[x][metric][:len(iters)]))
# for epoch in epochs:
# iters = log_dict[epoch]['iter']
# if log_dict[epoch]['mode'][-1] == 'val':
# iters = iters[:-1]
# xs.append(
# np.array(iters) + (epoch - 1) * num_iters_per_epoch)
# ys.append(np.array(log_dict[epoch][metric][:len(iters)]))
xs = np.concatenate(xs)
ys = np.concatenate(ys)
plt.xlabel('iter')
plt.plot(
xs, ys, label=legend[i * num_metrics + j], linewidth=0.5)
plt.legend()
if args.title is not None:
plt.title(args.title)
if args.out is None:
plt.show()
else:
print(f'save curve to: {
args.out}')
plt.savefig(args.out)
plt.cla()
7.过程文件
其中含有my_yolox_l.py，faster_rcnn_r50_fpn_1x_coco_20200130-047c8118.pth：
链接：https://pan.baidu.com/s/1rEqjj8kzFaRS_3ynXFX2MQ
提取码：4e34