pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple

pip install -i https://mirror.baidu.com/pypi/simple paddlehub

import paddlehub as hub

# 加载模型
yolov3 = hub.Module(name="yolov3_darknet53_coco2017")

# 图片位置
test_img_path = "imgs/c1.jpg"

# 输入图片
input_dict = {"image": [test_img_path]}

# 输出结果
results = yolov3.object_detection(data=input_dict)
for result in results:
    print(result['path'])
    print(result['data']) 

>> python predict.py
[{'left': 684.79376, 'right': 2024.4724, 'top': 961.53644, 'bottom': 2299.855, 'label': 'cat', 'confidence': 0.94765514}, {'left': 1461.0829, 'right': 3853.3633, 'top': 621.53064, 'bottom': 2769.5376, 'label': 'cat', 'confidence': 0.8093604}]

def paint_rect(input_img: str, output_path: str,
               labels: list, position: list):
    """
    画出矩形
        :param input_img: 输入图片
        :param output_path: 输出图片
        :param labels: 标签
        :param positions: 坐标
    """
    img = cv2.imread(input_img)

    for position in positions:
        print(position)
        # 画矩形框, 输入参数分别为图像、左上角坐标、右下角坐标、颜色数组、粗细
        cv2.rectangle(
            img, (position['left'], position['top']),
            (position['right'], position['bottom']),
            (0, 255, 0), thickness=10
        )

    if 'cat' in labels:
        # 若是猫，则存到另一个地方
        shutil.move(input_img, output_path + os.sep + input_img.split('/')[-1])
        cv2.imwrite(output_path + os.sep + 'rect_%s' % input_img.split('/')[-1], img) 

def get_all_path(dirpath, *suffix):
    """
    获得所有路径

    @param dirpath: 目录
    @param *suffix: 后缀
    """

    path_array = []
    for r, ds, fs in os.walk(dirpath):
        for fn in fs:
            if os.path.splitext(fn)[1] in suffix:
                fname = os.path.join(r, fn)
                path_array.append(fname)
    return path_array 
# 获得所有jpg和png图片
image_paths = get_all_path(source_path, '.jpg', '.JPG', 'png', 'PNG') 

# 加载模型
yolov3 = hub.Module(name="yolov3_darknet53_coco2017")
# 输入图片
input_dict = {"image": image_paths}
# 输出结果
results = yolov3.object_detection(data=input_dict, labels=['cat']) 

def paint_rect(input_img: str, output_path: str,
               labels: list, position: list):
    """
    画出矩形
        :param input_img: 输入图片
        :param output_path: 输出图片
        :param labels: 标签
        :param positions: 坐标
    """
    img = cv2.imread(input_img)

    for position in positions:
        # 画矩形框, 输入参数分别为图像、左上角坐标、右下角坐标、颜色数组、粗细
        cv2.rectangle(
            img, (position['left'], position['top']),
            (position['right'], position['bottom']),
            (0, 255, 0), thickness=10
        )

    if 'cat' in labels:
        # 若是猫，则存到另一个地方
        shutil.move(input_img, output_path + os.sep + input_img.split('/')[-1])
        cv2.imwrite(output_path + os.sep + 'rect_%s' % input_img.split('/')[-1], img) 

results = yolov3.object_detection(data=input_dict, labels=['cat'])
for result in results:
    path = result['path']
    labels = []
    positions = []
    for target in result['data']:
        labels.append(target.get('label', ''))
        positions.append({
            'left': target.get('left', -1),
            'top': target.get('top', -1),
            'right': target.get('right', -1),
            'bottom': target.get('bottom', -1)
        })
    paint_rect(path, target_path, labels, positions)  

import paddlehub as hub
import cv2
import os
import shutil


def get_all_path(dirpath, *suffix):
    """
    获得所有路径

    @param dirpath: 目录
    @param *suffix: 后缀
    """

    path_array = []
    for r, ds, fs in os.walk(dirpath):
        for fn in fs:
            if os.path.splitext(fn)[1] in suffix:
                fname = os.path.join(r, fn)
                path_array.append(fname)
    return path_array


def paint_rect(input_img: str, output_path: str,
               labels: list, position: list):
    """
    画出矩形
        :param input_img: 输入图片
        :param output_path: 输出图片
        :param labels: 标签
        :param positions: 坐标
    """
    img = cv2.imread(input_img)

    for position in positions:
        # 画矩形框, 输入参数分别为图像、左上角坐标、右下角坐标、颜色数组、粗细
        cv2.rectangle(
            img, (position['left'], position['top']),
            (position['right'], position['bottom']),
            (0, 255, 0), thickness=10
        )

    if 'cat' in labels:
        # 若是猫，则存到另一个地方
        shutil.move(input_img, output_path + os.sep + input_img.split('/')[-1])
        cv2.imwrite(output_path + os.sep + 'rect_%s' % input_img.split('/')[-1], img)


if __name__ == '__main__':
    source_path = './imgs/'
    target_path = './target/'

    # 获得所有jpg和png图片
    image_paths = get_all_path(source_path, '.jpg', '.JPG', 'png', 'PNG')

    # 加载模型
    yolov3 = hub.Module(name="yolov3_darknet53_coco2017")

    # 输入图片
    input_dict = {"image": image_paths}

    # 输出结果
    results = yolov3.object_detection(data=input_dict, labels=['cat'])
    for result in results:
        path = result['path']
        labels = []
        positions = []
        for target in result['data']:
            labels.append(target.get('label', ''))
            positions.append({
                'left': target.get('left', -1),
                'top': target.get('top', -1),
                'right': target.get('right', -1),
                'bottom': target.get('bottom', -1)
            })
        paint_rect(path, target_path, labels, positions)