import glob
import tensorflow as tf

from datasets.DetectionDataset import MapillaryLikeDetectionFileListDataset
from datasets.KITTI.segtrack.KITTI_MOTS_info import SEQ_IDS_TRAIN, SEQ_IDS_VAL
from datasets.Loader import register_dataset
from datasets.util.Util import username
from datasets import DataKeys
from shutil import copytree
from random import randint
from core.Log import log
import os

NAME = "KITTI_segtrack"
NAME_DETECTION = "KITTI_segtrack_detection"
DEFAULT_PATH = "/globalwork/" + username() + "/data/KITTI_MOTS/train/"
ID_DIVISOR = 1000
CLASS_IDS_WITH_INSTANCES = [1, 2]
CROWD_ID = 10
NUM_CLASSES = 3  # background, car, pedestrian
N_MAX_DETECTIONS = 100

用于检测单个图像

# used for detection on individual images
@register_dataset(NAME_DETECTION)
class KittiSegtrackDetectionDataset(MapillaryLikeDetectionFileListDataset):
  # 得到验证集seq_ids_train和训练集seq_ids_val的序号
  def __init__(self, config, subset, name=NAME, default_path=DEFAULT_PATH):
    self.seq_ids_train = SEQ_IDS_TRAIN
    self.seq_ids_val = SEQ_IDS_VAL
    self.imgs_are_pngs = config.bool("imgs_are_pngs", True)
    t = config.string_list("seq_ids_train", []) #得到训练集的序号
    if t:
      self.seq_ids_train = t
    v = config.string_list("seq_ids_val", []) #得到验证集的序号
    if v:
      self.seq_ids_val = v
    super().__init__(config, name, subset, default_path, NUM_CLASSES, N_MAX_DETECTIONS, CLASS_IDS_WITH_INSTANCES,
                     ID_DIVISOR, crowd_id=CROWD_ID)
    self.copy_dataset_to_tmp = config.bool("copy_dataset_to_tmp", False)
    if self.copy_dataset_to_tmp:
      print("Copying dataset to $TMP!", file=log.v1)
      new_path = "$TMP/" + str(randint(1, 100000))
      new_path = os.path.expandvars(new_path)
      os.makedirs(new_path)
      print("Copying images...", file=log.v1)
      copytree(self.data_dir + "/images", new_path + "/images")
      print("Copying instances...", file=log.v1)
      copytree(self.data_dir + "/instances", new_path + "/instances")
      self.data_dir = new_path

  def read_inputfile_lists(self):
    seq_ids = self.seq_ids_train if self.subset == "train" else self.seq_ids_val
    anns = []
    for seq_id in seq_ids:
      anns += sorted(glob.glob(self.data_dir + "/instances/" + seq_id + "/*.png"))
             # sorted:按升序返回包含迭代中所有项目的新列表。
             # 可以提供自定义关键字函数来自定义排序顺序，并且可以设置反向标志以得到降序结果。

                   # glob:返回与路径名称模式匹配的路径列表。
                   # 该模式可能包含简单的shell样式通配符la fnmatch。但是，与fnmatch不同，以点开头的文件名是特殊情况，与“ * ”和“？”不匹配模式。
                  # 如果递归为真，则模式'**'将匹配任何文件和零个或多个目录和子目录。
    imgs = [x.replace("/instances/", "/images/") for x in anns]  #将ann中所有/instances/替换为/images/
            #Python replace() 方法把字符串中的 old（旧字符串） 替换成 new(新字符串)，如果指定第三个参数max，则替换不超过 max 次。
    if not self.imgs_are_pngs: #如果图片不是png格式，那么替换为jpg格式，mots图片都是png格式
      imgs = [x.replace(".png", ".jpg") for x in imgs]
    return imgs, anns

• 不明白为什么要先匹配实例/instances/再替换为图像/images/，直接为图像/images/不行吗?