关键点数据集MTII

MTII数据集下载地址：http://human-pose.mpi-inf.mpg.de/#download
MPII人体姿势数据集，1.0版
马克斯·普朗克信息学院版权所有2015
MPII 样本数： 25000 个（单人、多人）
16个关键点，0 - r ankle, 1 - r knee, 2 - r hip,3 - l hip,4 - l knee, 5 - l ankle, 6 - l ankle， 7 - l ankle，8 - upper neck, 9 - head top,10 - r wrist,11 - r elbow, 12 - r shoulder, 13 - l shoulder,14 - l elbow, 15 - l wrist

Annotation description
Annotations are stored in a matlab structure RELEASE having following fields

.annolist(imgidx) - annotations for image imgidx
	.image.name - image filename
	.annorect(ridx) - body annotations for a person ridx
	.x1, .y1, .x2, .y2 - coordinates of the head rectangle
	.scale - person scale w.r.t. 200 px height
	.objpos - rough human position in the image
	.annopoints.point - person-centric body joint annotations
		.x, .y - coordinates of a joint
		.id - joint id (0 - r ankle, 1 - r knee, 2 - r hip, 3 - l hip, 4 - l knee, 5 - l ankle, 6 - pelvis, 7 - thorax, 8 				- upper neck, 9 - head top, 10 - r wrist, 11 - r elbow, 12 - r shoulder, 13 - l shoulder, 14 - l elbow, 15 - l wrist)
is_visible - joint visibility
	.vidx - video index in video_list
	.frame_sec - image position in video, in seconds
.img_train(imgidx) - training/testing image assignment
.single_person(imgidx) - contains rectangle id ridx of sufficiently separated individuals
.act(imgidx) - activity/category label for image imgidx
	.act_name - activity name
	.cat_name - category name
	.act_id - activity id
.video_list(videoidx) - specifies video id as is provided by YouTube. To watch video on youtube go to https://www.youtube.com/watch?v=video_list(videoidx)

我下载的数据集images大小为24984张图片数据，解析出的有关键点的标注文档个数为17209个txt文件（16个点以及不足16个点全部解析），和别人解析的数目对不上，具体原因没有找到

"""
摘录：https://github.com/Fangyh09/PoseDatasets
"""
from scipy.io import loadmat
import numpy as np


def save_joints():
    mat = loadmat('E:\迅雷下载\mpii_human_pose_v1_u12_2\\mpii_human_pose_v1_u12_1.mat')
    fout = open(r'E:\迅雷下载\mpii_human_pose_v1_u12_2\\mpii_list1.txt', 'w')

    for i, (anno, train_flag) in enumerate(
            zip(mat['RELEASE']['annolist'][0, 0][0],
                mat['RELEASE']['img_train'][0, 0][0],
                )):
        img_fn = anno['image']['name'][0, 0][0]
        train_flag = int(train_flag)

        if 'annopoints' in str(anno['annorect'].dtype):
            # only one person
            annopoints = anno['annorect']['annopoints'][0]
            head_x1s = anno['annorect']['x1'][0]
            head_y1s = anno['annorect']['y1'][0]
            head_x2s = anno['annorect']['x2'][0]
            head_y2s = anno['annorect']['y2'][0]


            for annopoint, head_x1, head_y1, head_x2, head_y2 in zip(
                    annopoints, head_x1s, head_y1s, head_x2s, head_y2s):
                if annopoint != []:
                    head_rect = [float(head_x1[0, 0]),
                                 float(head_y1[0, 0]),
                                 float(head_x2[0, 0]),
                                 float(head_y2[0, 0])]
                    feed_dict = {}
                    feed_dict['width'] = int(abs(float(head_x2[0, 0]) - float(head_x1[0, 0])))
                    feed_dict['height'] = int(abs(float(head_y2[0, 0]) - float(head_y1[0, 0])))
                    # joint coordinates
                    annopoint = annopoint['point'][0, 0]
                    j_id = [str(j_i[0, 0]) for j_i in annopoint['id'][0]]
                    x = [x[0, 0] for x in annopoint['x'][0]]
                    y = [y[0, 0] for y in annopoint['y'][0]]

                    joint_pos = {}
                    for _j_id, (_x, _y) in zip(j_id, zip(x, y)):
                        joint_pos[str(_j_id)] = [float(_x), float(_y)]


                    if 'is_visible' in str(annopoint.dtype):
                        vis = [v[0] if v else [0]
                               for v in annopoint['is_visible'][0]]
                        vis = dict([(k, int(v[0])) if len(v) > 0 else v
                                    for k, v in zip(j_id, vis)])
                    else:
                        vis = None
                    feed_dict['x'] = x
                    feed_dict['y'] = y
                    feed_dict['vis'] = vis
                    feed_dict['filename'] = img_fn

                    data = {
                            'filename': img_fn,
                            'train': train_flag,
                            'head_rect': head_rect,
                            'is_visible': vis,
                            'joint_pos': joint_pos
                        }
    #

                headlocation = ' '.join(str(x) for x in head_rect)
                label = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9', '10', '11', '12', '13', '14', '15']
                sss = ' '

                for key in label:
                    if key in data['joint_pos'].keys():
                        print("key",key)
                        sss = sss + str(key) + ' '  + str(int(data['joint_pos'][key][0])) + ' ' + str(int(data['joint_pos'][key][1])) + ' ' + str(int(data['is_visible'][key])) + ' '
                    else:
                        pass
                print(sss)
                sss = sss.strip()
                with open("E:\迅雷下载\mpii_human_pose_v1_u12_2\\txt1\\" + data['filename'].split(".")[0] + '.txt', 'a+') as f:
                    f.write(sss +";" + headlocation + '\n')
    #             fout.write(data['filename'] + ' ' + sss + '\n')
    # fout.close()


if __name__ == '__main__':
    save_joints()

可视化

# -*- coding: utf-8 -*-
"""
mat格式转成txt格式
"""
import numpy as np
import cv2
import os

if __name__ == '__main__':
    path1 = r"E:\mpii_human_pose_v1_u12_2\txt\\"#注意这里不要有中文路径，否则cv2.imread的时候无法读取数据
    path2 = r"E:\mpii_human_pose_v1_u12_2\images\\"
    path3 = r"E:\mpii_human_pose_v1_u12_2\vis\\"
    txtList = os.listdir(path1)
    imagesList = os.listdir(path2)
    for txt in txtList:
        txtPath = path1 + txt
        imgPath = path2 + txt.split('.')[0] + ".jpg"
        saveImgPath = path3 + txt.split('.')[0] + ".jpg"
        print(imgPath)
        img = cv2.imread(imgPath)
        with open(txtPath, 'r') as f:
            lines = f.readlines()
            for line in lines:
                pointLocation = line.split(';')[0].split(' ')
                print(pointLocation)
                for i in range(int(len(pointLocation) / 4)):
                    print(pointLocation[i * 4],pointLocation[i * 4 + 1],pointLocation[i * 4 + 2])
                    cv2.circle(img, (int(pointLocation[i * 4 + 1]), int(pointLocation[i * 4 + 2])), 1, (0, 0, 255), 2)
                    cv2.putText(img, pointLocation[i * 4], (int(pointLocation[i * 4 + 1]), int(pointLocation[i * 4 + 2])), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
                    #
            cv2.imwrite(saveImgPath, img)

            cv2.imshow("img", img)
            cv2.waitKey(1)
            # exit(0)

参考：https://blog.csdn.net/u013841196/article/details/107104183