pybaseutils 2.1.7__tar.gz → 2.2.1__tar.gz

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/PKG-INFO

@@ -1,12 +1,11 @@
 Metadata-Version: 2.1
 Name: pybaseutils
-Version: 2.1.7
+Version: 2.2.1
 Summary: pybaseutils
 Home-page: https://github.com/PanJinquan/base-utils
 Author: PanJinquan
 Author-email: 390737991@qq.com
 License: MIT
-Platform: UNKNOWN
 License-File: LICENCE
 
 base-utils
@@ -139,5 +138,3 @@ base_utils(C++)
        return 0;
    }
 
-
-

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/__init__.py

@@ -5,4 +5,4 @@
     @Date   : 2019-05-07 17:40:27
 """
 
-__version__ = '2.1.7'
+__version__ = '2.2.1'

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/coords_utils.py

@@ -267,7 +267,7 @@ def get_section(start, end, nums=2, scale=1.0, dtype=None):
     return out
 
 
-def box_iou_v1(box1, box2):
+def get_boxes_iou(box1, box2):
     """
     :param box1: 预测框(n, 4)
     :param box2: GT框 (m, 4)
@@ -294,6 +294,9 @@ def box_iou_v1(box1, box2):
     return iou
 
 
+box_iou_v1 = get_boxes_iou
+
+
 def box_iou_v2(box1, box2):
     """
     https://www.jb51.net/article/178732.htm
@@ -321,6 +324,64 @@ def box_iou_v2(box1, box2):
     return iou
 
 
+def get_box_iou(box1, box2):
+    """
+    计算IOU=交集(A,B)/并集(A,B)
+    计算IOM=交集(A,B)/最小集(A,B)
+    :param box1: = [xmin1, ymin1, xmax1, ymax1]
+    :param box2: = [xmin2, ymin2, xmax2, ymax2]
+    :return:
+    """
+    if len(box1) == 0: return 0
+    if len(box2) == 0: return 0
+    xmin1, ymin1, xmax1, ymax1 = box1
+    xmin2, ymin2, xmax2, ymax2 = box2
+    # 计算每个矩形的面积
+    s1 = (xmax1 - xmin1) * (ymax1 - ymin1)  # C的面积
+    s2 = (xmax2 - xmin2) * (ymax2 - ymin2)  # G的面积
+
+    # 计算相交矩形
+    xmin = max(xmin1, xmin2)
+    ymin = max(ymin1, ymin2)
+    xmax = min(xmax1, xmax2)
+    ymax = min(ymax1, ymax2)
+
+    w = max(0, xmax - xmin)
+    h = max(0, ymax - ymin)
+    area = w * h  # C∩G的面积
+    iou = area / (s1 + s2 - area)
+    return iou
+
+
+def get_box_iom(box1, box2):
+    """
+    计算IOU=交集(A,B)/并集(A,B)
+    计算IOM=交集(A,B)/最小集(A,B)
+    :param box1: = [xmin1, ymin1, xmax1, ymax1]
+    :param box2: = [xmin2, ymin2, xmax2, ymax2]
+    :return:
+    """
+    if len(box1) == 0: return 0
+    if len(box2) == 0: return 0
+    xmin1, ymin1, xmax1, ymax1 = box1
+    xmin2, ymin2, xmax2, ymax2 = box2
+    # 计算每个矩形的面积
+    s1 = (xmax1 - xmin1) * (ymax1 - ymin1)  # C的面积
+    s2 = (xmax2 - xmin2) * (ymax2 - ymin2)  # G的面积
+
+    # 计算相交矩形
+    xmin = max(xmin1, xmin2)
+    ymin = max(ymin1, ymin2)
+    xmax = min(xmax1, xmax2)
+    ymax = min(ymax1, ymax2)
+
+    w = max(0, xmax - xmin)
+    h = max(0, ymax - ymin)
+    area = w * h  # C∩G的面积
+    iou = area / min(s1, s2)
+    return iou
+
+
 class YOLOCoords(object):
     def __init__(self, max_boxes=120, norm=False):
         self.max_boxes = max_boxes

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/cvutils/video_utils.py

@@ -245,11 +245,11 @@ def video_capture(video_file: int or str, save_video: str or int = None, interva
     end = min(end, num_frames)  # TODO 当num_frames<0时，使用0<end<count继续播放
     interval = fps if interval == -1 else interval  # 当interval=-1，表示interval=fps,即一秒一帧
     save_fps = max(kwargs.get("speed", 1) * fps // interval, 1)
-    count = start
+    count = 0
     video_writer = None
     while True:
         ret, frame = video_cap.read()
-        if count % interval == 0 and count >= 0:
+        if count % interval == 0 and count >= start:
             # TODO 设置抽帧的位置，但某些格式视频容易出现问题
             # if isinstance(video_file, str): video_cap.set(cv2.CAP_PROP_POS_FRAMES, count)
             # ret, frame = video_cap.read()
@@ -297,11 +297,11 @@ def video_iterator(video_file: int or str, save_video: str or int = None, interv
     end = min(end, num_frames)  # TODO 当num_frames<0时，使用0<end<count继续播放
     interval = fps if interval == -1 else interval  # 当interval=-1，表示interval=fps,即一秒一帧
     save_fps = max(kwargs.get("speed", 1) * fps // interval, 1)
-    count = start
+    count = 0
     video_writer = None
     while True:
         ret, frame = video_cap.read()
-        if count % interval == 0 and count >= 0:
+        if count % interval == 0 and count >= start:
             # TODO 设置抽帧的位置，但某些格式视频容易出现问题
             # if isinstance(video_file, str): video_cap.set(cv2.CAP_PROP_POS_FRAMES, count)
             # ret, frame = video_cap.read()

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/dataloader/base_dataset.py

@@ -123,6 +123,10 @@ class Dataset(object):
                 image_id.append(line)
         return image_id
 
+    @staticmethod
+    def read_files(filename, split=None):  # TODO 避免兼容问题
+        return Dataset.read_file(filename, split=split)
+
     @staticmethod
     def get_file_list(dir_root, postfix=['*.jpg'], basename=True):
         """

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/dataloader/parser_image_folder.py

@@ -26,6 +26,7 @@ class FolderDataset(parser_image_text.TextDataset):
         :param shuffle:
         :param disp:
         """
+        self.image_dir = image_dir
         super(FolderDataset, self).__init__(data_file=image_dir,
                                             data_root=None,
                                             class_name=class_name,

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/dataloader/parser_image_text.py

@@ -38,10 +38,14 @@ class TextDataset(Dataset):
         :param check:
         :param kwargs: use_max,use_mean,crop_scale,resample,save_info
         """
-        self.kwargs = kwargs
+        self.data_file = data_file
+        self.data_root = data_root
         self.use_rgb = use_rgb
         self.transform = transform
         self.phase = phase
+        self.shuffle = shuffle
+        self.check = check
+        self.kwargs = kwargs
         self.label_index = kwargs.get("label_index", "label")  # 类别字段key
         self.class_name, self.class_dict = self.parser_classes(class_name)
         self.item_list = self.parser_dataset(data_file, data_root=data_root, label_index=self.label_index,

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/dataloader/parser_labelme.py

@@ -391,6 +391,7 @@ def show_target_image(image, bboxes, labels, points, color=(), thickness=1):
     # image = image_utils.draw_key_point_in_image(image, points)
     image = image_utils.draw_image_contours(image, points, labels, color=color, thickness=thickness)
     image_utils.cv_show_image("det", image)
+    return image
 
 
 if __name__ == "__main__":

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/file_utils.py

@@ -25,6 +25,7 @@ from tqdm import tqdm
 
 IMG_POSTFIX = ['*.jpg', '*.jpeg', '*.png', '*.tif', "*.JPG", "*.bmp"]
 VIDEO_POSTFIX = ['*.mp4', '*.avi', '*.mov', "*.flv"]
+AUDIO_POSTFIX = ['*.mp3', '*.wav']
 
 
 def str2bool(v):
@@ -71,6 +72,21 @@ def get_file_size(file, rate=1024):
     return os.path.getsize(file) / rate
 
 
+def replace_elements(items, src, dst, ignore=True):
+    """
+    将列表中，值为src改为dst
+    :param items:
+    :param src:
+    :param dst:
+    :param ignore: 是否提出空的数据
+    :return:
+    """
+    for i, item in enumerate(items):
+        if item == src: items[i] = dst
+    if ignore: items = [it for it in items if str(it)]
+    return items
+
+
 def combine_flags(flags: list, use_time=True, info=True):
     """
     :param flags:

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/image_utils.py

@@ -631,6 +631,7 @@ def resize_image_padding(image, size, color=(0, 0, 0), interpolation=cv2.INTER_L
     :param color: 短边进行填充的color value
     :return:
     """
+    if not size: return image
     h, w = image.shape[:2]
     if w / h > size[0] / size[1] > 0:
         dsize = (size[0], None)
@@ -1753,8 +1754,6 @@ def nms_boxes_cv2(boxes: np.ndarray, scores: np.ndarray, labels: np.ndarray, sco
     return index
 
 
-
-
 def file2base64(file):
     image_base64 = base64.b64encode(open(file, 'rb').read()).decode()
     return image_base64
@@ -2880,15 +2879,15 @@ def frames2gif_by_pil(frames, gif_file="test.gif", fps=2, loop=0, use_rgb=False)
                        optimize=False, loop=loop)
 
 
-def image_dir2gif(image_dir, size=None, gif_file=None, interval=1, fps=4, loop=0, padding=True, use_pil=True):
+def image_dir2gif(image_dir, size=None, gif_file=None, interval=1, fps=4, loop=0, padding=False, crop=[], use_pil=True):
     if not gif_file: gif_file = os.path.join(os.path.dirname(image_dir), os.path.basename(image_dir) + ".gif")
     file_list = file_utils.get_images_list(image_dir)
     image_file2gif(file_list, size=size, gif_file=gif_file, interval=interval, fps=fps, loop=loop,
-                   padding=padding, use_pil=use_pil)
+                   padding=padding, crop=crop, use_pil=use_pil)
 
 
 def image_file2gif(file_list, size=None, gif_file="test.gif", interval=1, fps=4, loop=0,
-                   padding=True, use_pil=False):
+                   padding=True, crop=[], use_pil=False):
     """
     pip install imageio
     uri：合成后的gif动图的名字，可以随意更改。
@@ -2902,6 +2901,7 @@ def image_file2gif(file_list, size=None, gif_file="test.gif", interval=1, fps=4,
     :param size:gif图片的大小
     :param gif_file: 输出的GIF图的路径
     :param fps: 刷新频率
+    :param crop: 裁剪(xmin,ymin,xmax,ymax)
     :param loop: 循环次数
     :param use_pil: True使用PIL库生成GIF图，文件小，但质量较差
                     False使用imageio库生成GIF图，文件大，但质量较好
@@ -2912,6 +2912,7 @@ def image_file2gif(file_list, size=None, gif_file="test.gif", interval=1, fps=4,
         if count % interval == 0:
             bgr = cv2.imread(file)
             image = cv2.cvtColor(bgr, cv2.COLOR_BGR2RGB)
+            if len(crop) == 4: image = get_box_crop(image, box=crop)
             if padding:
                 image = resize_image_padding(image, size=size)
             else:

pybaseutils-2.2.1/pybaseutils/log.py

@@ -0,0 +1,79 @@
+# -*-coding: utf-8 -*-
+"""
+    @Author : PKing
+    @E-mail : 
+    @Date   : 2024-10-27 13:42:12
+    @Brief  :
+"""
+import os
+import sys
+from loguru import logger
+
+LOG_FORMAT = {
+    "simple":   "{time:YYYY-MM-DD HH:mm:ss}|{level:7}| {message}",
+    "name":     "{time:YYYY-MM-DD HH:mm:ss}|{level:7}|{name} {line:4}| {message}",  # 打印文件名
+    "module":   "{time:YYYY-MM-DD HH:mm:ss}|{level:7}|{name}.{module} {line:4}| {message}",  # 打印模块名
+    "function": "{time:YYYY-MM-DD HH:mm:ss}|{level:7}|{name}.{module}.{function} {line:4}| {message}",  # 打印函数
+}
+
+
+def set_logger(name="", level="debug", logfile=None, format="simple", is_main_process=True):
+    """
+    logger = set_logger(level="debug", logfile="log.txt")
+    url: https://www.cnblogs.com/shiyitongxue/p/17870527.html
+    :param level: 设置log输出级别:debug,info,warning,error
+    :param logfile: log保存路径，如果为None，则在控制台打印log
+    :param is_main_process: 是否是主进程
+    :return:
+    """
+    format = LOG_FORMAT.get(format, LOG_FORMAT.get("simple"))
+    if is_main_process:
+        logger.configure(handlers=[{"sink": sys.stderr,  # 打印到控制台，sys.stderr表示控制台
+                                    "format": format, "colorize": True, "level": level.upper()},
+                                   {"sink": logfile,  # 输出到文件，文件名app.log
+                                    "format": format, "colorize": False, "level": level.upper(), "rotation": "100 MB",
+                                    "retention": "10 days"}
+                                   ])
+    else:
+        logger.configure(handlers=[{"sink": sys.stderr,  # 打印到控制台，sys.stderr表示控制台
+                                    "format": format, "colorize": True, "level": "ERROR"}
+                                   ])
+    return logger
+
+
+def set_logger_v2(name="", level="debug", logfile=None, format=None, is_main_process=True):
+    """
+    logger = set_logger(level="debug", logfile="log.txt")
+    url: https://www.cnblogs.com/shiyitongxue/p/17870527.html
+    :param level: 设置log输出级别:debug,info,warning,error
+    :param logfile: log保存路径，如果为None，则在控制台打印log
+    :param is_main_process: 是否是主进程
+    :return:
+    """
+    format = LOG_FORMAT.get(format, LOG_FORMAT.get("line"))
+    logger.remove(0)  # 去除默认的LOG
+    if is_main_process:
+        # 每天创建一个新的文件，一个星期定期清理一次
+        logger.add(logfile, level=level.upper(), rotation="1 day", retention="7 days", format=format)
+        logger.add(sys.stderr, level=level.upper(), format=format)
+    else:
+        logger.add(sys.stderr, level="ERROR", format=format)
+    return logger
+
+
+def get_logger():
+    return logger
+
+
+def example():
+    logger = get_logger()
+    logger.debug("debug")
+    logger.info("info")
+    logger.warning("warning")
+    logger.error("error")
+
+
+if __name__ == '__main__':
+    logfile = "./log.log"
+    logger = set_logger(logfile=logfile, is_main_process=True, level="debug")
+    example()

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils/numpy_utils.py

@@ -13,7 +13,7 @@ import heapq
 
 def feature_norm(x, axis=-1):
     """
-    特征归一化:
+    特征归一化,会使得特征L2之和为1, 即||y||=L2(y,axis=-1)
     y = x / x.norm(dim=-1, keepdim=True)               # torch
     y = x / np.linalg.norm(x, axis=-1, keepdims=True)  # numpy
     :param x: 输入二维矩阵(N,embedding-size)，每行是一个样本，样本特征维度为embedding-size
@@ -24,6 +24,21 @@ def feature_norm(x, axis=-1):
     return y
 
 
+def torch_norm(x, axis=-1):
+    """
+    特征归一化:
+    y = x / x.norm(dim=-1, keepdim=True)               # torch
+    y = x / np.linalg.norm(x, axis=-1, keepdims=True)  # numpy
+    :param x: 输入二维矩阵(N,embedding-size)，每行是一个样本，样本特征维度为embedding-size
+    :param axis:
+    :return:
+    """
+    import torch
+    x = torch.from_numpy(x)
+    y = x / x.norm(dim=-1, keepdim=True)  # torch
+    return y
+
+
 def feature_similarity(inputs, target):
     """
     计算特征相似性
@@ -645,8 +660,11 @@ class Preprocessing(object):
 if __name__ == "__main__":
     from pybaseutils import numpy_utils
 
-    y = np.array([[0., 0.1, 0], [0, 0.1, 1], [0, 0.1, 2]])
-    x = np.array([[0, 0.1, 1], [0, 0.1, 2], [0., 0.1, 0]])
-    x = feature_norm(x)
+    y = np.array([[0., 0.1, 0, 1], [0, 0.1, 1, 1], [0, 0.1, 2, 1]])
+    x = np.array([[0, 0.1, 1, 1], [0, 0.1, 2, 1], [0., 0.1, 0, 1]])
+    x1 = feature_norm(x)
+    x2 = torch_norm(x).numpy()
+    print(l2(x1, axis=-1))
+    print(l2(x2, axis=-1))
     y = feature_norm(y)
-    print(feature_matching(x, y))
+    print(feature_matching(x1, y))

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils.egg-info/PKG-INFO

@@ -1,12 +1,11 @@
 Metadata-Version: 2.1
 Name: pybaseutils
-Version: 2.1.7
+Version: 2.2.1
 Summary: pybaseutils
 Home-page: https://github.com/PanJinquan/base-utils
 Author: PanJinquan
 Author-email: 390737991@qq.com
 License: MIT
-Platform: UNKNOWN
 License-File: LICENCE
 
 base-utils
@@ -139,5 +138,3 @@ base_utils(C++)
        return 0;
    }
 
-
-

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/pybaseutils.egg-info/SOURCES.txt

@@ -14,6 +14,7 @@ pybaseutils/heatmap_utils.py
 pybaseutils/image_utils.py
 pybaseutils/json_utils.py
 pybaseutils/log.py
+pybaseutils/log_utils.py
 pybaseutils/logger.py
 pybaseutils/numpy_utils.py
 pybaseutils/pandas_utils.py
@@ -132,6 +133,7 @@ test_py/demo_ffmpy.py
 test_py/demo_for_pair_file.py
 test_py/demo_for_polygon.py
 test_py/demo_for_trt.py
+test_py/demo_get_file_label.py
 test_py/demo_get_file_list.py
 test_py/demo_gif.py
 test_py/demo_gif_video.py
@@ -150,6 +152,7 @@ test_py/demo_standard_image .py
 test_py/demo_standard_video .py
 test_py/demo_taichi.py
 test_py/demo_video.py
+test_py/demo_video_crop.py
 test_py/demo_voc_crop.py
 test_py/demo_voc_vis.py
 test_py/demo_word_similar.py
@@ -159,6 +162,8 @@ test_py/get_file_list.py
 test_py/kafka_worker.py
 test_py/men_tracemalloc.py
 test_py/performance.py
+test_py/Image_enhance/__init__.py
+test_py/Image_enhance/dmeo01.py
 test_py/WebCrawler/__init__.py
 test_py/WebCrawler/search_image.py
 test_py/WebCrawler/search_image_for_baidu.py
@@ -238,6 +243,11 @@ test_py/image_correction/__init__.py
 test_py/image_correction/demo_correction_v1.py
 test_py/image_correction/demo_correction_v2.py
 test_py/image_correction/demo_correction_v3.py
+test_py/milvus_demo/__init__.py
+test_py/milvus_demo/demo01.py
+test_py/milvus_demo/demo02.py
+test_py/milvus_demo/hello_milvus.py
+test_py/milvus_demo/milvus_client.py
 test_py/pose/__init__.py
 test_py/pose/human_pose.py
 test_py/redis_py/__init__.py

pybaseutils-2.2.1/test_py/Image_enhance/__init__.py

@@ -0,0 +1,7 @@
+# -*-coding: utf-8 -*-
+"""
+    @Author : PKing
+    @E-mail : 
+    @Date   : 2024-10-25 19:36:07
+    @Brief  :
+"""

pybaseutils-2.2.1/test_py/Image_enhance/dmeo01.py

@@ -0,0 +1,31 @@
+# -*-coding: utf-8 -*-
+"""
+    @Author : PKing
+    @E-mail : 390737991@qq.com
+    @Date   : 2022-12-31 11:37:30
+    @Brief  : https://blog.csdn.net/qdPython/article/details/121381363
+"""
+import cv2
+import random
+import types
+import numpy as np
+from typing import Callable
+from pybaseutils import image_utils, file_utils, text_utils
+from pybaseutils.cvutils import video_utils
+import cv2
+import re
+
+
+def equalizeHist(image):
+    image = cv2.equalizeHist(image)
+    return image
+
+
+if __name__ == '__main__':
+    image_dir = "/home/PKing/nasdata/dataset-dmai/AIJE/暗光数据"
+    image_list = file_utils.get_images_list(image_dir)
+    for image_file in image_list:
+        src = cv2.imread(image_file)
+        dst = equalizeHist(src)
+        image_utils.cv_show_image("dst", dst, delay=10)
+        image_utils.cv_show_image("src", src, delay=0)

pybaseutils-2.2.1/test_py/aije/convert_labelme2voc.py

@@ -0,0 +1,47 @@
+# -*-coding: utf-8 -*-
+"""
+    @Author : Pan
+    @E-mail : 390737991@qq.com
+    @Date   : 2022-10-12 16:21:55
+    @Brief  :
+"""
+import sys
+import os
+
+sys.path.insert(0, os.getcwd())
+import cv2
+from pybaseutils.converter.convert_labelme2voc import Labelme2VOC
+
+
+def build_dataset(anno_dir, class_dict=None):
+    # anno_dir = "/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v1-det/dataset-v01/json"
+    data_root = os.path.dirname(anno_dir)
+    image_dir = os.path.join(data_root, "JPEGImages")
+    # class_dict = {"身穿工作服": "person", "未穿工作服": "person"}
+    # out_root = os.path.join(data_root, "person")
+    # out_image_dir = os.path.join(data_root, "person")
+    print(anno_dir)
+    lm = Labelme2VOC(image_dir, anno_dir, class_name=class_dict, shuffle=False)
+    lm.build_dataset(data_root, class_dict={}, out_img=False, vis=False, crop=False)
+
+
+def build_datasets():
+    anno_dirs = [
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v21/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v22/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v23/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v24/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v25/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v26/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v27/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v28/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v29/json',
+        '/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v29-test/json',
+    ]
+    for anno_dir in anno_dirs:
+        build_dataset(anno_dir)
+
+
+if __name__ == "__main__":
+    # 将AIJE项目数据集，转换为VOC数据集
+    build_datasets()

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/test_py/aije/demo_labelme_crop.py

@@ -64,7 +64,7 @@ if __name__ == "__main__":
           '绝缘手套', '绝缘鞋', '脚扣', '螺丝', '身穿工作服', '遮拦杆', '铁架', '铝扎线']
     室外: []
     """
-    anno_dir = "/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-v2-det/dataset-v21/json"
+    anno_dir = "/home/PKing/nasdata/dataset-dmai/AIJE/dataset/aije-action-cvlm-v2/test-v1/aije-action-train-v02/json"
     class_name = None  # 室内\
     out_dir = os.path.join(os.path.dirname(anno_dir), "crops")
     dataset = parser_labelme.LabelMeDatasets(filename=None,
@@ -77,7 +77,7 @@ if __name__ == "__main__":
                                              shuffle=False)
     print("have num:{}".format(len(dataset)))
     class_name = dataset.class_name
-    scale = [1.0, 1.0]
+    scale = [1.5,1.5]
     flag = str(scale[0]).replace(".", "p")
     flag = None
     # scale = None

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/test_py/aije/video_convertor.py

@@ -16,12 +16,13 @@ from pybaseutils import image_utils, file_utils
 def video_converter(inp_dir, out_dir):
     files = file_utils.get_files_list(inp_dir, postfix=["*.dav"])
     for video_file in tqdm(files):
+        print(video_file)
         save_video = file_utils.create_dir(out_dir, None, os.path.basename(video_file))
         save_video = save_video.replace(".dav", ".mp4")
         video_utils.video2video(video_file, save_video, interval=1, vis=False)
 
 
 if __name__ == '__main__':
-    inp_dir = "/home/PKing/nasdata/dataset-dmai/AIJE/技能人才系统_数据集管理/江门四维数据/江门四维2024-08-06/NVR录像文件"
-    out_dir = "/home/PKing/nasdata/dataset-dmai/AIJE/技能人才系统_数据集管理/江门四维数据/江门四维2024-08-06/MP4录像文件"
+    inp_dir = "/home/PKing/Downloads/NVR"
+    out_dir = "/home/PKing/Downloads/NVR"
     video_converter(inp_dir, out_dir)

pybaseutils-2.2.1/test_py/demo1.py

@@ -0,0 +1,55 @@
+# -*-coding: utf-8 -*-
+"""
+    @Author : PKing
+    @E-mail :
+    @Date   : 2024-02-05 18:19:18
+    @Brief  :
+"""
+import os
+import cv2
+import numpy as np
+import random
+import json
+from PIL import Image
+from typing import Dict
+from tqdm import tqdm
+from pybaseutils import image_utils, file_utils, json_utils, base64_utils, time_utils
+from pybaseutils.cvutils import video_utils
+from pybaseutils.transforms import transform_utils
+from pybaseutils.converter import build_voc, build_labelme
+from pybaseutils.dataloader import parser_labelme
+import xmltodict
+
+import torch
+import torch.nn as nn
+
+
+class CMDLoss(nn.Module):
+    """计算两个分布的中心矩差差异的Loss函数"""
+
+    def __init__(self, num_channels, use_gpu=True):
+        super(CMDLoss, self).__init__()
+        self.use_gpu = use_gpu
+        self.num_channels = num_channels
+        self.register_buffer('one', torch.tensor(1.0))
+
+    def forward(self, input, target):
+        # 计算均值
+        input_mean = torch.mean(input, dim=[2, 3], keepdim=True)
+        target_mean = torch.mean(target, dim=[2, 3], keepdim=True)
+
+        # 计算方差
+        input_var = torch.mean((input - input_mean) ** 2, dim=[2, 3], keepdim=True)
+        target_var = torch.mean((target - target_mean) ** 2, dim=[2, 3], keepdim=True)
+
+        # 计算协方差
+        input_mean_expand = input_mean.expand_as(input)
+        target_mean_expand = target_mean.expand_as(target)
+        cov = torch.mean((input - input_mean_expand) * (target - target_mean_expand), dim=[2, 3], keepdim=True)
+
+        # 计算CMD
+        c = torch.mean(self.one / self.num_channels * cov, dim=[2, 3], keepdim=True)
+        d = torch.mean(self.one / self.num_channels * (input_var - target_var), dim=[2, 3], keepdim=True)
+        cmd_loss = c / (d + 1e-5)
+
+        return cmd_loss

{pybaseutils-2.1.7 → pybaseutils-2.2.1}/test_py/demo2.py

@@ -8,6 +8,7 @@
 import cv2
 import random
 import types
+import torch
 import numpy as np
 from typing import Callable
 from pybaseutils import image_utils, file_utils, text_utils
@@ -16,9 +17,7 @@ import cv2
 import re
 
 if __name__ == '__main__':
-    # 示例使用
-    subs = ["拿", '手*']  #
-    texts = ["拿着", "手拿安全帽", "手拿绝缘手套", "手拿工具袋"]
-    # 使用通配符查找子串
-    matches = text_utils.find_match_texts(texts, subs, org=True)
-    print(matches)
+    accs = np.asarray([0.6205, 0.5752, 0.9296, 0.9906, 0.4833, 0.0756, 0.000001, 0, 0, 0.8205, 0.9205])
+    print(accs,np.mean(accs))
+    accs = accs[accs > 0]
+    print(accs,np.mean(accs))