foreword
I am learning mmdetect recently. Generally speaking, the framework of mmlab feels quite difficult to get started. I also combined the up masters of station b (the name of the up master: I am Tudui, OneShotLove, more expensive than flying birds_HKL ) and Zhihu The mmlab official watch and learn, it is really a nanny-level tutorial, highly recommended. But in order to prevent forgetting in the future, record it. If there is something wrong, everyone is welcome to criticize and correct.
mmdetect2d trains its own data set (2) - model training
1. Data set preparation
Labelme is used for data set labeling. There are many tutorials on the use of labelme on the Internet, and they are also very detailed, so I won’t go into details. The final obtained file should have a .png (or .jpg) and a corresponding .json file (annotation content), and the picture and the corresponding annotation name should be the same. As shown in the figure below: 
If you didn’t pay attention when marking, you can write a piece of code and rename it:
import os
import re
# 运行时将该文件放到和json文件同一目录下，或者直接修改dir_path路径
dir_path = "./"
# 定义正则化规则
formula = re.compile('"imagePath": "(.+?png)",')
for file in os.listdir(dir_path):
# 判定是否是json文件
if os.path.splitext(file)[-1] != '.json':
continue
# 读取json文件
with open(os.path.join(dir_path, file), encoding='utf-8') as f:
content = f.read()
# 根据正则规则找到json文件里的imagePath
imagePath = formula.findall(content)[0]
print('imagePath ', imagePath)
new_content = content.replace(imagePath, os.path.splitext(file)[0] + '.png')
#重命名并保存
with open(os.path.join(dir_path, file), 'w', encoding='utf-8') as nf:
nf.write(new_content)
The ultimate goal is to ensure that the name of the corresponding json file is consistent with the picture before data processing.
2. The structure of the coco data set:
The folder structure tree corresponding to the coco dataset is as follows:
Among them, train2017 stores training set images:
val2017 stores validation set images:
annotation folder stores corresponding label json files:
3. Generate a standard coco dataset format
The json file generated by labelme corresponds to one image, but the coco data set is collected, so it needs to be written into the same json file. In addition, there is no distinction between the training set and the verification set when labeling, so when generating data, it needs to be divided and adjusted according to the ratio you need. code show as below:
import os
import json
import numpy as np
import glob
import cv2
from sklearn.model_selection import train_test_split
from tqdm import tqdm
from labelme import utils
# 背景值为0，因此标签对应的mask值应从1开始，根据自己需要识别的种类来。
classname_to_id = {
"hole": 1,
}
class Lableme2CoCo:
def __init__(self):
self.images = []
self.annotations = []
self.categories = []
self.img_id = 0
self.ann_id = 0
# 定义写入json文件的函数
def save_coco_json(self, instance, save_path):
json.dump(instance, open(save_path, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
# 由json文件构建COCO
def to_coco(self, json_path_list):
self._init_categories()
for json_path in json_path_list:
obj = self.read_jsonfile(json_path)
self.images.append(self._image(obj, json_path))
shapes = obj['shapes']
for shape in shapes:
annotation = self._annotation(shape)
self.annotations.append(annotation)
self.ann_id += 1
self.img_id += 1
instance = {
}
instance['info'] = 'spytensor created'
instance['license'] = ['license']
instance['images'] = self.images
instance['annotations'] = self.annotations
instance['categories'] = self.categories
return instance
# 构建类别
def _init_categories(self):
for k, v in classname_to_id.items():
category = {
}
category['id'] = v
category['name'] = k
self.categories.append(category)
# 构建COCO的image字段
def _image(self, obj, path):
image = {
}
img_x = utils.img_b64_to_arr(obj['imageData'])
h, w = img_x.shape[:-1]
image['height'] = h
image['width'] = w
image['id'] = self.img_id
image['file_name'] = os.path.basename(path).replace(".json", ".jpg")
return image
# 构建COCO的annotation字段
def _annotation(self, shape):
# print('shape', shape)
label = shape['label']
points = shape['points']
annotation = {
}
annotation['id'] = self.ann_id
annotation['image_id'] = self.img_id
annotation['category_id'] = int(classname_to_id[label])
annotation['segmentation'] = [np.asarray(points).flatten().tolist()]
annotation['bbox'] = self._get_box(points)
annotation['iscrowd'] = 0
annotation['area'] = 1.0
return annotation
# 读取json文件，返回一个json对象
def read_jsonfile(self, path):
with open(path, "r", encoding='utf-8') as f:
return json.load(f)
# COCO的格式： [x1,y1,w,h] 对应COCO的bbox格式
def _get_box(self, points):
min_x = min_y = np.inf
max_x = max_y = 0
for x, y in points:
min_x = min(min_x, x)
min_y = min(min_y, y)
max_x = max(max_x, x)
max_y = max(max_y, y)
return [min_x, min_y, max_x - min_x, max_y - min_y]
if __name__ == '__main__':
# 训练前准备
# labelme文件路径
labelme_path = r"F:\mywork\Semantic-Segmentation-master\SegNet_Mobile\data\mydata"
# 储存文件路径
saved_coco_path = r"F:\mywork\Semantic-Segmentation-master\SegNet_Mobile\data\mydatacoco"
# annotation文件路径
annotation_path = os.path.join(saved_coco_path, 'annotations')
# train_img文件路径
train_img_path = os.path.join(saved_coco_path, 'train2017')
# val_img文件路径
val_img_path = os.path.join(saved_coco_path, 'val2017')
print('reading...')
# 创建文件夹
if not os.path.exists(annotation_path):
os.makedirs(annotation_path)
if not os.path.exists(train_img_path):
os.makedirs(train_img_path)
if not os.path.exists(val_img_path):
os.makedirs(val_img_path)
# 获取images目录下所有的json文件列表
print(labelme_path + "/*.json")
json_list_path = glob.glob(labelme_path + "/*.json")
print('json_list_path: ', len(json_list_path))
# 划分训练集和验证集,这里没有区分val2017和tran2017目录，所有图片都放在images目录下
# 验证集和训练集比例为1：9，可以根据自己需要更改
train_path, val_path = train_test_split(json_list_path, test_size=0.1, train_size=0.9)
print("train_n:", len(train_path), 'val_n:', len(val_path))
# 把训练集转化为COCO的json格式
# 将训练集对应annotation的json文件写入同一个
l2c_train = Lableme2CoCo()
train_instance = l2c_train.to_coco(train_path)
l2c_train.save_coco_json(train_instance, '%s/annotations/instances_train2017.json' % saved_coco_path)
# 将训练对应的图片放入train2017文件夹
for file in tqdm(train_path):
# 根据划分的json文件来索引训练集
img_name = file.replace('json', 'png')
temp_img = cv2.imread(img_name)
try:
# 分离文件名与路径
(filepath, tempfilename) = os.path.split(img_name)
# 将文件改为jpg格式，因为coco数据集中都是使用jpg格式
tmp_new_name = tempfilename.replace('png', 'jpg')
# 获取新文件的保存路径
train_img = os.path.join(train_img_path, tmp_new_name)
# 写入图片
cv2.imwrite(train_img, temp_img)
except Exception as e:
print(e)
print('Wrong Image:', img_name)
continue
# 把验证集转化为COCO的json格式
# 将验证集对应annotation的json文件写入同一个
l2c_val = Lableme2CoCo()
val_instance = l2c_val.to_coco(val_path)
l2c_val.save_coco_json(val_instance, '%s/annotations/instances_val2017.json' % saved_coco_path)
# 将验证集对应的图片放入val2017文件夹
for file in tqdm(val_path):
# shutil.copy(file.replace("json", "jpg"), "%scoco/images/val2017/" % saved_coco_path)
img_name = file.replace('json', 'png')
temp_img = cv2.imread(img_name)
try:
# 分离文件名与路径
(filepath, tempfilename) = os.path.split(img_name)
# 将文件改为jpg格式，因为coco数据集中都是使用jpg格式
tmp_new_name = tempfilename.replace('png', 'jpg')
# 获取新文件的保存路径
val_img = os.path.join(val_img_path, tmp_new_name)
# 写入图片
cv2.imwrite(val_img, temp_img)
except Exception as e:
print(e)
print('Wrong Image:', img_name)
continue
At this point, the data processing ends.