基于OpenPose的人体姿态检测(非常好)

参考:https://blog.csdn.net/yph001/article/details/83218839

一、概述

OpenPose最开始由卡内基梅隆大学提出,其主要基于先后发表的几篇文章中提出的模型中进行实现:
CVPR 2016: Convolutional Pose Machine(CPM)
CVPR2017 : realtime multi-person pose estimation
CVPR2017 : Hand Keypoint Detection in Single Images using Multiview Bootstrapping
但运行计算量非常大,通常得在GPU上运行,并且帧率较低(低于5fps),在此后也陆续出现了一些改进版
改进版主要在模型上进行了一些改进或裁剪,另外移动端(如各种尬舞app) 为能够跑通OpenPose,在改网络结构的同时,对算法本身也进行了优化,减少了计算量,但与此同时准确性也有相应地降低。

二、简化版OpenPose实现代码

  • 代码来源GitHub:human-pose-estimation-opencv
  • 其代码较为简单,模型(较小:7.8M)已经训练好在graph_opt.pb文件中,其中全部实现代码在openpose.py文件中,下面是实现代码及测试效果:
# To use Inference Engine backend, specify location of plugins:
# export LD_LIBRARY_PATH=/opt/intel/deeplearning_deploymenttoolkit/deployment_tools/external/mklml_lnx/lib:$LD_LIBRARY_PATH
import cv2 as cv
import numpy as np
import argparse

parser = argparse.ArgumentParser()
parser.add_argument('--input', help='Path to image or video. Skip to capture frames from camera')
parser.add_argument('--thr', default=0.2, type=float, help='Threshold value for pose parts heat map')
parser.add_argument('--width', default=368, type=int, help='Resize input to specific width.')
parser.add_argument('--height', default=368, type=int, help='Resize input to specific height.')

args = parser.parse_args()

BODY_PARTS = { "Nose": 0, "Neck": 1, "RShoulder": 2, "RElbow": 3, "RWrist": 4,
               "LShoulder": 5, "LElbow": 6, "LWrist": 7, "RHip": 8, "RKnee": 9,
               "RAnkle": 10, "LHip": 11, "LKnee": 12, "LAnkle": 13, "REye": 14,
               "LEye": 15, "REar": 16, "LEar": 17, "Background": 18 }

POSE_PAIRS = [ ["Neck", "RShoulder"], ["Neck", "LShoulder"], ["RShoulder", "RElbow"],
               ["RElbow", "RWrist"], ["LShoulder", "LElbow"], ["LElbow", "LWrist"],
               ["Neck", "RHip"], ["RHip", "RKnee"], ["RKnee", "RAnkle"], ["Neck", "LHip"],
               ["LHip", "LKnee"], ["LKnee", "LAnkle"], ["Neck", "Nose"], ["Nose", "REye"],
               ["REye", "REar"], ["Nose", "LEye"], ["LEye", "LEar"] ]

inWidth = args.width
inHeight = args.height

net = cv.dnn.readNetFromTensorflow("graph_opt.pb")

cap = cv.VideoCapture(args.input if args.input else 0)

while cv.waitKey(1) < 0:
    hasFrame, frame = cap.read()
    if not hasFrame:
        cv.waitKey()
        break

    frameWidth = frame.shape[1]
    frameHeight = frame.shape[0]
    
    net.setInput(cv.dnn.blobFromImage(frame, 1.0, (inWidth, inHeight), (127.5, 127.5, 127.5), swapRB=True, crop=False))
    out = net.forward()
    out = out[:, :19, :, :]  # MobileNet output [1, 57, -1, -1], we only need the first 19 elements

    assert(len(BODY_PARTS) == out.shape[1])

    points = []
    for i in range(len(BODY_PARTS)):
        # Slice heatmap of corresponging body's part.
        heatMap = out[0, i, :, :]

        # Originally, we try to find all the local maximums. To simplify a sample
        # we just find a global one. However only a single pose at the same time
        # could be detected this way.
        _, conf, _, point = cv.minMaxLoc(heatMap)
        x = (frameWidth * point[0]) / out.shape[3]
        y = (frameHeight * point[1]) / out.shape[2]
        # Add a point if it's confidence is higher than threshold.
        points.append((int(x), int(y)) if conf > args.thr else None)

    for pair in POSE_PAIRS:
        partFrom = pair[0]
        partTo = pair[1]
        assert(partFrom in BODY_PARTS)
        assert(partTo in BODY_PARTS)

        idFrom = BODY_PARTS[partFrom]
        idTo = BODY_PARTS[partTo]

        if points[idFrom] and points[idTo]:
            cv.line(frame, points[idFrom], points[idTo], (0, 255, 0), 3)
            cv.ellipse(frame, points[idFrom], (3, 3), 0, 0, 360, (0, 0, 255), cv.FILLED)
            cv.ellipse(frame, points[idTo], (3, 3), 0, 0, 360, (0, 0, 255), cv.FILLED)

    t, _ = net.getPerfProfile()
    freq = cv.getTickFrequency() / 1000
    cv.putText(frame, '%.2fms' % (t / freq), (10, 20), cv.FONT_HERSHEY_SIMPLEX, 0.5, (0, 0, 0))

    cv.imshow('OpenPose using OpenCV', frame)

检测效果如下:
下面两张图中体现出单人且姿态展开时较好的检测效果:

下面两张图中体现出多人或特殊姿态时较差的检测效果(如乱连接或者遗漏关键点等):
基于OpenPose的人体姿态检测(非常好)_第1张图片
从左上角显示的处理时间可看到,处理较慢,基本一张图片需耗时0.5S

较复杂版OpenPose实现代码

  • 代码来源GitHub:camera-openpose-keras
  • 其代码比起前面这个更复杂一些,模型(更大:200M)已经训练好在可自行下载,但在外网不易下载,因此也可在百度云下载:model.h5
  • 其中全部实现代码在demo_camera.py文件中,下面是修改了一点点代码,采取读入图片的方式进行了测试:
import argparse
import cv2
import math
import time
import numpy as np
import util
from config_reader import config_reader
from scipy.ndimage.filters import gaussian_filter
from model import get_testing_model

tic=0
# find connection in the specified sequence, center 29 is in the position 15
limbSeq = [[2, 3], [2, 6], [3, 4], [4, 5], [6, 7], [7, 8], [2, 9], [9, 10], \
           [10, 11], [2, 12], [12, 13], [13, 14], [2, 1], [1, 15], [15, 17], \
           [1, 16], [16, 18], [3, 17], [6, 18]]

# the middle joints heatmap correpondence
mapIdx = [[31, 32], [39, 40], [33, 34], [35, 36], [41, 42], [43, 44], [19, 20], [21, 22], \
          [23, 24], [25, 26], [27, 28], [29, 30], [47, 48], [49, 50], [53, 54], [51, 52], \
          [55, 56], [37, 38], [45, 46]]

# visualize
colors = [[255, 0, 0], [255, 85, 0], [255, 170, 0], [255, 255, 0], [170, 255, 0], [85, 255, 0],
          [0, 255, 0], \
          [0, 255, 85], [0, 255, 170], [0, 255, 255], [0, 170, 255], [0, 85, 255], [0, 0, 255],
          [85, 0, 255], \
          [170, 0, 255], [255, 0, 255], [255, 0, 170], [255, 0, 85]]


def process (input_image, params, model_params):

    oriImg = cv2.imread(input_image)  # B,G,R order
    multiplier = [x * model_params['boxsize'] / oriImg.shape[0] for x in params['scale_search']]

    heatmap_avg = np.zeros((oriImg.shape[0], oriImg.shape[1], 19))
    paf_avg = np.zeros((oriImg.shape[0], oriImg.shape[1], 38))

    #for m in range(len(multiplier)):
    for m in range(1):
        scale = multiplier[m]

        imageToTest = cv2.resize(oriImg, (0, 0), fx=scale, fy=scale, interpolation=cv2.INTER_CUBIC)
        imageToTest_padded, pad = util.padRightDownCorner(imageToTest, model_params['stride'],
                                                          model_params['padValue'])

        input_img = np.transpose(np.float32(imageToTest_padded[:,:,:,np.newaxis]), (3,0,1,2)) # required shape (1, width, height, channels)

        
        output_blobs = model.predict(input_img)
      
        
        
        # extract outputs, resize, and remove padding
        heatmap = np.squeeze(output_blobs[1])  # output 1 is heatmaps
        heatmap = cv2.resize(heatmap, (0, 0), fx=model_params['stride'], fy=model_params['stride'],
                             interpolation=cv2.INTER_CUBIC)
        heatmap = heatmap[:imageToTest_padded.shape[0] - pad[2], :imageToTest_padded.shape[1] - pad[3],
                  :]
        heatmap = cv2.resize(heatmap, (oriImg.shape[1], oriImg.shape[0]), interpolation=cv2.INTER_CUBIC)

        paf = np.squeeze(output_blobs[0])  # output 0 is PAFs
        paf = cv2.resize(paf, (0, 0), fx=model_params['stride'], fy=model_params['stride'],
                         interpolation=cv2.INTER_CUBIC)
        paf = paf[:imageToTest_padded.shape[0] - pad[2], :imageToTest_padded.shape[1] - pad[3], :]
        paf = cv2.resize(paf, (oriImg.shape[1], oriImg.shape[0]), interpolation=cv2.INTER_CUBIC)

        heatmap_avg = heatmap_avg + heatmap / len(multiplier)
        paf_avg = paf_avg + paf / len(multiplier)

    all_peaks = []
    peak_counter = 0
    prinfTick(1)
    for part in range(18):
        map_ori = heatmap_avg[:, :, part]
        map = gaussian_filter(map_ori, sigma=3)

        map_left = np.zeros(map.shape)
        map_left[1:, :] = map[:-1, :]
        map_right = np.zeros(map.shape)
        map_right[:-1, :] = map[1:, :]
        map_up = np.zeros(map.shape)
        map_up[:, 1:] = map[:, :-1]
        map_down = np.zeros(map.shape)
        map_down[:, :-1] = map[:, 1:]

        peaks_binary = np.logical_and.reduce(
            (map >= map_left, map >= map_right, map >= map_up, map >= map_down, map > params['thre1']))
        peaks = list(zip(np.nonzero(peaks_binary)[1], np.nonzero(peaks_binary)[0]))  # note reverse
        peaks_with_score = [x + (map_ori[x[1], x[0]],) for x in peaks]
        id = range(peak_counter, peak_counter + len(peaks))
        peaks_with_score_and_id = [peaks_with_score[i] + (id[i],) for i in range(len(id))]

        all_peaks.append(peaks_with_score_and_id)
        peak_counter += len(peaks)

    connection_all = []
    special_k = []
    mid_num = 10
    prinfTick(2)
    for k in range(len(mapIdx)):
        score_mid = paf_avg[:, :, [x - 19 for x in mapIdx[k]]]
        candA = all_peaks[limbSeq[k][0] - 1]
        candB = all_peaks[limbSeq[k][1] - 1]
        nA = len(candA)
        nB = len(candB)
        indexA, indexB = limbSeq[k]
        if (nA != 0 and nB != 0):
            connection_candidate = []
            for i in range(nA):
                for j in range(nB):
                    vec = np.subtract(candB[j][:2], candA[i][:2])
                    norm = math.sqrt(vec[0] * vec[0] + vec[1] * vec[1])
                    # failure case when 2 body parts overlaps
                    if norm == 0:
                        continue
                    vec = np.divide(vec, norm)

                    startend = list(zip(np.linspace(candA[i][0], candB[j][0], num=mid_num), \
                                   np.linspace(candA[i][1], candB[j][1], num=mid_num)))

                    vec_x = np.array(
                        [score_mid[int(round(startend[I][1])), int(round(startend[I][0])), 0] \
                         for I in range(len(startend))])
                    vec_y = np.array(
                        [score_mid[int(round(startend[I][1])), int(round(startend[I][0])), 1] \
                         for I in range(len(startend))])

                    score_midpts = np.multiply(vec_x, vec[0]) + np.multiply(vec_y, vec[1])
                    score_with_dist_prior = sum(score_midpts) / len(score_midpts) + min(
                        0.5 * oriImg.shape[0] / norm - 1, 0)
                    criterion1 = len(np.nonzero(score_midpts > params['thre2'])[0]) > 0.8 * len(
                        score_midpts)
                    criterion2 = score_with_dist_prior > 0
                    if criterion1 and criterion2:
                        connection_candidate.append([i, j, score_with_dist_prior,
                                                     score_with_dist_prior + candA[i][2] + candB[j][2]])

            connection_candidate = sorted(connection_candidate, key=lambda x: x[2], reverse=True)
            connection = np.zeros((0, 5))
            for c in range(len(connection_candidate)):
                i, j, s = connection_candidate[c][0:3]
                if (i not in connection[:, 3] and j not in connection[:, 4]):
                    connection = np.vstack([connection, [candA[i][3], candB[j][3], s, i, j]])
                    if (len(connection) >= min(nA, nB)):
                        break

            connection_all.append(connection)
        else:
            special_k.append(k)
            connection_all.append([])

    # last number in each row is the total parts number of that person
    # the second last number in each row is the score of the overall configuration
    subset = -1 * np.ones((0, 20))
    candidate = np.array([item for sublist in all_peaks for item in sublist])
    prinfTick(3)
    for k in range(len(mapIdx)):
        if k not in special_k:
            partAs = connection_all[k][:, 0]
            partBs = connection_all[k][:, 1]
            indexA, indexB = np.array(limbSeq[k]) - 1

            for i in range(len(connection_all[k])):  # = 1:size(temp,1)
                found = 0
                subset_idx = [-1, -1]
                for j in range(len(subset)):  # 1:size(subset,1):
                    if subset[j][indexA] == partAs[i] or subset[j][indexB] == partBs[i]:
                        subset_idx[found] = j
                        found += 1

                if found == 1:
                    j = subset_idx[0]
                    if (subset[j][indexB] != partBs[i]):
                        subset[j][indexB] = partBs[i]
                        subset[j][-1] += 1
                        subset[j][-2] += candidate[partBs[i].astype(int), 2] + connection_all[k][i][2]
                elif found == 2:  # if found 2 and disjoint, merge them
                    j1, j2 = subset_idx
                    membership = ((subset[j1] >= 0).astype(int) + (subset[j2] >= 0).astype(int))[:-2]
                    if len(np.nonzero(membership == 2)[0]) == 0:  # merge
                        subset[j1][:-2] += (subset[j2][:-2] + 1)
                        subset[j1][-2:] += subset[j2][-2:]
                        subset[j1][-2] += connection_all[k][i][2]
                        subset = np.delete(subset, j2, 0)
                    else:  # as like found == 1
                        subset[j1][indexB] = partBs[i]
                        subset[j1][-1] += 1
                        subset[j1][-2] += candidate[partBs[i].astype(int), 2] + connection_all[k][i][2]

                # if find no partA in the subset, create a new subset
                elif not found and k < 17:
                    row = -1 * np.ones(20)
                    row[indexA] = partAs[i]
                    row[indexB] = partBs[i]
                    row[-1] = 2
                    row[-2] = sum(candidate[connection_all[k][i, :2].astype(int), 2]) + \
                              connection_all[k][i][2]
                    subset = np.vstack([subset, row])

    # delete some rows of subset which has few parts occur
    deleteIdx = [];
    for i in range(len(subset)):
        if subset[i][-1] < 4 or subset[i][-2] / subset[i][-1] < 0.4:
            deleteIdx.append(i)
    subset = np.delete(subset, deleteIdx, axis=0)

    canvas = cv2.imread(input_image)  # B,G,R order
    for i in range(18):
        for j in range(len(all_peaks[i])):
            cv2.circle(canvas, all_peaks[i][j][0:2], 4, colors[i], thickness=-1)

    stickwidth = 4
    for i in range(17):
        for n in range(len(subset)):
            index = subset[n][np.array(limbSeq[i]) - 1]
            if -1 in index:
                continue
            cur_canvas = canvas.copy()
            Y = candidate[index.astype(int), 0]
            X = candidate[index.astype(int), 1]
            mX = np.mean(X)
            mY = np.mean(Y)
            length = ((X[0] - X[1]) ** 2 + (Y[0] - Y[1]) ** 2) ** 0.5
            angle = math.degrees(math.atan2(X[0] - X[1], Y[0] - Y[1]))
            polygon = cv2.ellipse2Poly((int(mY), int(mX)), (int(length / 2), stickwidth), int(angle), 0,
                                       360, 1)
            cv2.fillConvexPoly(cur_canvas, polygon, colors[i])
            canvas = cv2.addWeighted(canvas, 0.4, cur_canvas, 0.6, 0)

    return canvas

def prinfTick(i):
    toc = time.time()
    print ('processing time%d is %.5f' % (i,toc - tic))        

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--image', type=str, default='sample_images/ski.jpg', help='input image')
    parser.add_argument('--output', type=str, default='result.png', help='output image')
    parser.add_argument('--model', type=str, default='model/keras/model.h5', help='path to the weights file')

    args = parser.parse_args()
    input_image = args.image
    output = args.output
    keras_weights_file = args.model

    tic = time.time()
    print('start processing...')

    # load model

    # authors of original model don't use
    # vgg normalization (subtracting mean) on input images
    model = get_testing_model()
    model.load_weights(keras_weights_file)
    
    cap=cv2.VideoCapture(0)
    vi=cap.isOpened()
    if(vi == False):
        time.sleep(2) #必须要此步骤,否则失败
        #fr = cv2.imread('./sample_images/ski2.jpg',1)
        tic = time.time()
        # cv2.imwrite(input_image, fr)
        params, model_params = config_reader()
        canvas = process(input_image, params, model_params)    
        cv2.imshow("capture",canvas)
        #cv2.waitKey(0)#修改使得按ESC能退出终止程序
        key = cv2.waitKey(0)
        if key == 27:
            cv2.destroyAllWindows()



    if(vi == True):
        cap.set(3,160)
        cap.set(4,120)
        time.sleep(2) #必须要此步骤,否则失败
    
        while(1):
            tic = time.time()
        
            ret,frame=cap.read()
            cv2.imwrite(input_image, frame)
            params, model_params = config_reader()
            
        # generate image with body parts
            canvas = process(input_image, params, model_params)    
            cv2.imshow("capture",canvas) 
            if cv2.waitKey(1) & 0xFF==ord('q'):
                break
        cap.release()   
cv2.destroyAllWindows()    

程序的运行方式如下:

python demo_camera.py --image ./sample_images/ski.jpg

-image后面接的是输入图片的路径
检测效果如下

首先对比了上面简化版代码效果不好的两张图在这个模型中的处理效果
基于OpenPose的人体姿态检测(非常好)_第2张图片
另外,结合其他图片测试了一下效果,发现人能够呈现清楚时,关节点能够比较准确得检测出来
基于OpenPose的人体姿态检测(非常好)_第3张图片
可能会遇到的问题:
from tensorflow.python.eager.context import get_config报错
参考:https://blog.csdn.net/qq_21237549/article/details/121280047

你可能感兴趣的