windows10复现DEtectionTRansformers（DETR）并实现⾃⼰的数据集DEtection TRansformers（DETR）
DEtection TRansformer（DETR）是Facebook AI的研究者提出的Transformer的视觉版本，⽤于⽬标检测和全景分割。这是第⼀个将Transformer成功整合为检测pipeline中⼼构建块的⽬标检测框架。
第⼀步，先将代码下载下来，然后在pycharm中打开，运⾏terminal，输⼊
pip install -
1. 打开anconda的命令界⾯输⼊activate your_env_name (激活anaconda虚拟环境)；
2. 进⼊coco源码setup.py所在⽬录 cocoapi-master\PythonAPI；
3. 运⾏python setup.py build_ext install即可安装完成。
解决了这个问题，安装其他的库就没得什么问题了，如果遇到vc++2015之类的问题建议在vs studio中把c++那⼀项也安装好。
第⼆步，将它的pth⽂件改⼀下，因为他是⽤的coco数据集，⽽我们只需要训练⾃⼰的数据集，就是下图这个⽂件
运⾏⼀下代码，就会⽣成⼀个你数据集所需要的物体数⽬的pth，记得改那个数字。
import torch
pretrained_weights = torch.load('detr-r50-e632da11.pth')
num_class =3 #这⾥是你的物体数+1，因为背景也算⼀个
pretrained_weights["model"]["class_embed.weight"].resize_(num_class+1,256)
getsavefilenamepretrained_weights["model"]["class_embed.bias"].resize_(num_class+1)
torch.save(pretrained_weights,"detr-r50_%d.pth"%num_class
运⾏完后会⽣成下图的⽂件：
第三步，准备⾃⼰的训练集，⾸先你要⽤⾃⼰的数据标记⼀个VOC类型的数据集，这⾥就不多赘述了，
可以⾃⼰百度下载labelimg来打标，然后将⽂件放⼊这⼏个⽂件夹，就可以了，然后就是转换成json格式的数据。
第四步，转换成json格式，⽣成的⽂件夹记得改为instances_train2017,json这种样⼦，如下图，
下⾯是voc转换为json的代码，⾃⼰更改⽂件路径
# coding:utf-8
# pip install lxml
import os
import glob
import json
import shutil
import numpy as np
import numpy as np
ElementTree as ET
path2 ="C:/Users/Desktop/VOC2007"
START_BOUNDING_BOX_ID =1
def get(root, name):
return root.findall(name)
def get_and_check(root, name, length):
vars = root.findall(name)
if len(vars)==0:
raise NotImplementedError('Can not find %s in%s.' %(name, root.tag))
if length >0and len(vars)!= length:
raise NotImplementedError('The size of %s is supposed to be %d, but is%d.' %(name, length,len(vars))) if length ==1:
vars = vars[0]
return vars
def convert(xml_list, json_file):
json_dict ={"images":[],"type":"instances","annotations":[],"categories":[]}
categories = pre_py()
bnd_id = START_BOUNDING_BOX_ID
all_categories ={}
for index, line in enumerate(xml_list):
# print("Processing %s"%(line))
xml_f = line
tree = ET.parse(xml_f)
root = t()
filename = os.path.basename(xml_f)[:-4]+".jpg"
image_id =20190000001+ index
size = get_and_check(root, 'size', 1)
width =int(get_and_check(size, 'width',1).text)
height =int(get_and_check(size, 'height',1).text)
image ={'file_name': filename, 'height': height, 'width': width, 'id': image_id}
json_dict['images'].append(image)
## Cruuently we do not support segmentation
# segmented = get_and_check(root, 'segmented', 1).text
# assert segmented == '0'
for obj in get(root, 'object'):
category =get_and_check(obj, 'name',1).text
if category in all_categories:
all_categories[category]+=1
else:
all_categories[category]=1
if category not in categories:
if only_care_pre_define_categories:
continue
new_id =len(categories)+1
print(
"[warning] category '{}' not in 'pre_define_categories'({}), create new id: {} automatically".format( category, pre_define_categories, new_id))
categories[category]= new_id
category_id = categories[category]
bndbox = get_and_check(obj, 'bndbox', 1)
xmin =int(float(get_and_check(bndbox, 'xmin',1).text))
ymin =int(float(get_and_check(bndbox, 'ymin',1).text))
xmax =int(float(get_and_check(bndbox, 'xmax',1).text))
ymax =int(float(get_and_check(bndbox, 'ymax',1).text))
assert (xmax > xmin),"xmax <= xmin, {}".format(line)
assert (ymax > ymin),"ymax <= ymin, {}".format(line)
assert (ymax > ymin),"ymax <= ymin, {}".format(line)
o_width =abs(xmax - xmin)
o_height =abs(ymax - ymin)
ann ={'area': o_width * o_height, 'iscrowd':0, 'image_id':
image_id, 'bbox':[xmin, ymin, o_width, o_height],
'category_id': category_id, 'id': bnd_id, 'ignore':0,
'segmentation':[]}
json_dict['annotations'].append(ann)
bnd_id = bnd_id +1
for cate, cid in categories.items():
cat ={'supercategory': 'none', 'id': cid, 'name': cate}
json_dict['categories'].append(cat)
json_fp =open(json_file,'w')
json_str = json.dumps(json_dict)
json_fp.write(json_str)
json_fp.close()
print("------------create {} done--------------".format(json_file))
print("find {} categories: {} -->>> your pre_define_categories {}: {}".format(len(all_categories), all_categories.keys(),
len(pre_define_categories),
pre_define_categories.keys()))
print("category: id --> {}".format(categories))
print(categories.keys())
print(categories.values())
if __name__ == '__main__':
classes =['bicycle', 'pottedplant', 'tvmonitor']
pre_define_categories ={}
for i, cls in enumerate(classes):
pre_define_categories[cls]= i +1
# pre_define_categories = {'a1': 1, 'a3': 2, 'a6': 3, 'a9': 4, "a10":5}
only_care_pre_define_categories = True
# only_care_pre_define_categories = False
train_ratio =0.9
save_json_train = 'instances_train2014.json'
save_json_val = 'instances_val2014.json'
xml_dir ="./tmp_xml"
xml_list = glob.glob(xml_dir +"/*.xml")
xml_list = np.sort(xml_list)
np.random.seed(100)
np.random.shuffle(xml_list)
train_num =int(len(xml_list)* train_ratio)
xml_list_train = xml_list[:train_num]
xml_list_val = xml_list[train_num:]
convert(xml_list_train, save_json_train)
convert(xml_list_val, save_json_val)
if ists(path2 +"/annotations"):
<(path2 +"/annotations")
os.makedirs(path2 +"/annotations")
if ists(path2 +"/images/train2014"):
<(path2 +"/images/train2014")
os.makedirs(path2 +"/images/train2014")
if ists(path2 +"/images/val2014"):
<(path2 +"/images/val2014")
os.makedirs(path2 +"/images/val2014")
f1 =open("","w")
for xml in xml_list_train:
img = xml[:-4]+".jpg"
img = xml[:-4]+".jpg"
f1.write(os.path.basename(xml)[:-4]+"\n")
f2 =open("","w")
for xml in xml_list_val:
img = xml[:-4]+".jpg"
f2.write(os.path.basename(xml)[:-4]+"\n")
f1.close()
f2.close()
print("-------------------------------")
print("train number:",len(xml_list_train))
print("val number:",len(xml_list_val))
这下⾯是将xml中的图⽚从⽂件夹选出来，⽤得到的可以⽤。
#将图⽚根据xml中的⽂件名挑选出来
from PIL import Image
from PIL import ImageEnhance
import os
import cv2
import numpy as np
def convert(input_dir1, input_dir2,output_dir):
for filename in os.listdir(input_dir1):
for filename1 in os.listdir(input_dir2):
path = input_dir1 +"/"+ filename#图⽚路径
path1=input_dir2+'/'+filename1#xml路径
#print(" ", path)
#print(" ", path1)
if path1[-10:-4]==path[-10:-4]:#只有图⽚代号和xml代号相等的时候才会保存
print(path[-10:-4])
image = Image.open(path)
image.save(output_dir +"/"+ filename[:-4]+".jpg")
if __name__ == '__main__':
input_dir1 ="C:/Users/Desktop/VOC2007_COCO/images"#输⼊图⽚路径
input_dir2="C:/Users/Desktop/xml/xml_train"#输⼊标注路径
output_dir ="C:/Users/Desktop/xml/train1"#保存路径
convert(input_dir1,input_dir2,output_dir)
python main.py --dataset_file "coco"--coco_path data/coco --epochs 100--lr=1e-4--batch_size=2--num_workers=4--output_dir="outputs"--resume="detr -r50_3.pth"
训练完后会在outputs⽣成下图的⽂件，log⽂件是记录每⼀个epoch的⼀些信息
到这⾥就完成了整个训练。
最后，就是拿⾃⼰的训练的模型进⾏测试，更改这⾥的图⽚路径为要测试的图⽚的路径，还有第19⾏的CLASSES=[]，记得改成⾃⼰的类别！
整个流程就结束了，这个代码还提供了画图的功能，在util中的plot_utils.py⽂件
在这个py⽂件下加⼊这个代码，路径⾃⼰改哦！
if __name__ == '__main__':
files =list(Path('../outputs/eval').glob('*.pth'))
plot_precision_recall(files)
plt.show()
plot_logs(logs=Path('D:/detr/outputs/log/'),fields=('class_error', 'loss_bbox_unscaled', 'mAP'), ewm_col=0, log_name='')
plt.show()
出现这个错误：
OMP: Error #15: Initializing libiomp5md.dll, but found libiomp5md.dll already initialized.
解决⽅法为
import os
就可解决问题
if __name__ == '__main__':
files =list(Path('../outputs/eval').glob('*.pth'))
plot_precision_recall(files)
plt.show()
plot_logs(logs=Path('D:/BaiduNetdiskDownload/detr/outputs/log/'),fields=('class_error', 'loss_bbox_unscaled', 'mAP'), ewm_col=0, log_name='') plt.show()
好了，到这⾥所有的步骤与我遇到的坑都告诉⼤家了，有什么问题可以留⾔！
版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。
发表评论