sort多目标跟踪代码解读
算法整体流程：
1.读取每一帧检测的结果det。其中det.txt如下：第一个数代表帧号，第三个数~第六个数代表目标(x,y,w,h)，第七个数代表得分score,其它数据不详。
1,-1,500,158,30.979,70.299,93.673,-3.70694,-7.16689,0
1,-1,246,218,40.258,91.355,69.358,-11.4773,-5.53043,0
1,-1,648,238,36.706,83.294,55.955,-8.82797,-12.7447,0
每一帧取出的数据为：
x, y, w, h, s
500, 158, 30.979, 70.299, 93.673
246, 218, 40.258, 91.355, 69.358
648, 238, 36.706, 83.294, 55.955
2.sort循环更新
1）对于上一帧的跟踪器，这一帧首先各做一次预测
2）将预测结果异常的跟踪器存入to_del数组
3）屏蔽出现无效数值的跟踪器，然后做了压缩？
4）逆向删除异常的跟踪器
5）将检测结果分配到跟踪器上，这里会得到三个list
跟踪结果与检测结果匹配上的、未匹配上的检测结果、未匹配上的跟踪器
6）对于匹配上的，将检测结果更新跟踪器；
对于未匹配上的检测结果，认为是新检测出来的目标，给他送入新的跟踪器，每天加一个新的跟踪器，他的id+1。第一帧全部是未匹配到的。
7）获取跟踪器的结果
from __future__ import print_function
from numba import jit
import os.path
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as patches
from skimage import io
from sklearn.utils.linear_assignment_ import linear_assignment
import glob
import time
import argparse
from filterpy.kalman import KalmanFilter
@jit
def iou(bb_test,bb_gt):
"""
Computes IUO between two bboxes in the form [x1,y1,x2,y2]
"""
#大的左上角的点x
xx1 = np.maximum(bb_test[0], bb_gt[0])
# 大的左上角的点y
yy1 = np.maximum(bb_test[1], bb_gt[1])
# 小的右下角点x
xx2 = np.minimum(bb_test[2], bb_gt[2])
# 小的右下角点y
yy2 = np.minimum(bb_test[3], bb_gt[3])
#相交部分的区域
w = np.maximum(0., xx2 - xx1)
h = np.maximum(0., yy2 - yy1)
#相交区域的面积
wh = w * h
#iou, = a/(s1+s2-a),a是相交的面积，s1是第一个Box面积，s2是第二个box面积
o = wh / ((bb_test[2]-bb_test[0])*(bb_test[3]-bb_test[1])
+ (bb_gt[2]-bb_gt[0])*(bb_gt[3]-bb_gt[1]) - wh)
return(o)
def convert_bbox_to_z(bbox):
"""
Takes a bounding box in the form [x1,y1,x2,y2] and returns z in the form
[x,y,s,r] where x,y is the centre of the box and s is the scale/area and r is
the aspect ratio
"""
w = bbox[2]-bbox[0]
h = bbox[3]-bbox[1]
x = bbox[0]+w/2.
y = bbox[1]+h/2.
s = w*h #面积
r = w/float(h)
#转成4行一列
return np.array([x,y,s,r]).reshape((4,1))
def convert_x_to_bbox(x,score=None):
"""
Takes a bounding box in the centre form [x,y,s,r] and returns it in the form
[x1,y1,x2,y2] where x1,y1 is the top left and x2,y2 is the bottom right
"""
w = np.sqrt(x[2]*x[3])
h = x[2]/w
if(score==None):
return np.array([x[0]-w/2.,x[1]-h/2.,x[0]+w/2.,x[1]+h/2.]).reshape((1,4))
else:
return np.array([x[0]-w/2.,x[1]-h/2.,x[0]+w/2.,x[1]+h/2.,score]).reshape((1,5))
class KalmanBoxTracker(object):
"""
This class represents the internel state of individual tracked objects observed as bbox.
"""
count = 0
def __init__(self,bbox):
"""
Initialises a tracker using initial bounding box.
"""
#define constant velocity model
self.kf = KalmanFilter(dim_x=7, dim_z=4)
self.kf.F = np.array([[1,0,0,0,1,0,0],[0,1,0,0,0,1,0],[0,0,1,0,0,0,1],[0,0,0,1,0,0,0], [0,0,0,0,1,0,0],[0,0,0,0,0,1,0],[0,0,0,0,0,0,1]])
self.kf.H = np.array([[1,0,0,0,0,0,0],[0,1,0,0,0,0,0],[0,0,1,0,0,0,0],[0,0,0,1,0,0,0]])
self.kf.R[2:,2:] *= 10.
self.kf.P[4:,4:] *= 1000. #give high uncertainty to the unobservable initial velocities
self.kf.P *= 10.
self.kf.Q[-1,-1] *= 0.01
self.kf.Q[4:,4:] *= 0.01
self.kf.x[:4] = convert_bbox_to_z(bbox)
self.time_since_update = 0
self.id = KalmanBoxTracker.count
KalmanBoxTracker.count += 1
self.history = []
self.hits = 0
self.hit_streak = 0
self.age = 0
def update(self,bbox):
"""
Updates the state vector with observed bbox.
"""
self.time_since_update = 0
self.history = []
self.hits += 1
self.hit_streak += 1
self.kf.update(convert_bbox_to_z(bbox))
def predict(self):
"""
Advances the state vector and returns the predicted bounding box estimate.
"""
if((self.kf.x[6]+self.kf.x[2])<=0):
self.kf.x[6] *= 0.0
self.kf.predict()
self.age += 1
if(self.time_since_update>0):
self.hit_streak = 0
self.time_since_update += 1
self.history.append(convert_x_to_bbox(self.kf.x))
return self.history[-1]
def get_state(self):
"""
Returns the current bounding box estimate.
"""
return convert_x_to_bbox(self.kf.x)
def associate_detections_to_trackers(detections,trackers,iou_threshold = 0.3):
"""
将检测结果指定给跟踪目标
返回匹配到、检测未匹配到、跟踪未匹配到
"""
#如果跟踪器的目标个数为0
if(len(trackers)==0):
return np.empty((0,2),dtype=int), np.arange(len(detections)), np.empty((0,5),dtype=int)
#iou矩阵，检测的目标框和跟踪的目标框的iou
iou_matrix = np.zeros((len(detections),len(trackers)),dtype=np.float32)
#对于每个检测的结果，计算它和每个跟踪器结果的Iou
for d,det in enumerate(detections):
for t,trk in enumerate(trackers):
iou_matrix[d,t] = iou(det,trk)
#这里指的是利用匈牙利算法匹配跟踪和检测的iou
matched_indices = linear_assignment(-iou_matrix)
print("matched_indices: ", matched_indices)
'''
matched_indices:
[[0 1]
[1 0]
[2 2]]
'''
unmatched_detections = []
#对于没有匹配到的检测结果，将它存放在unmatched_detections
for d,det in enumerate(detections):
if(d not in matched_indices[:,0]):#matched_indices[:,0]代表匹配的检测器，matched_indices[:,1]匹配的跟踪器
unmatched_detections.append(d)#将没匹配到检测编号存起来
unmatched_trackers = []
for t,trk in enumerate(trackers):
if(t not in matched_indices[:,1]):
unmatched_trackers.append(t)
#低IOU匹配滤除
matches = []
for m in matched_indices:
#如果iou小于阈值的话，将检测和跟踪分别存入未匹配检测、未匹配跟踪器中
if(iou_matrix[m[0],m[1]]<iou_threshold):
unmatched_detections.append(m[0])
unmatched_trackers.append(m[1])
else:
matches.append(m.reshape(1,2))#转换成1行2列[检测器， 跟踪器]
if(len(matches)==0):
matches = np.empty((0,2),dtype=int)
else:
matches = np.concatenate(matches,axis=0)
return matches, np.array(unmatched_detections), np.array(unmatched_trackers)
class Sort(object):
def __init__(self,max_age=1,min_hits=3):
"""
Sets key parameters for SORT
"""
self.max_age = max_age
self.min_hits = min_hits
self.trackers = []
self.frame_count = 0
def update(self,dets):
"""
Params:
dets是一组数组(x1,y1,x2,y2,score)
dets - a numpy array of detections in the format [[x1,y1,x2,y2,score],[x1,y1,x2,y2,score],...]
Requires: this method must be called once for each frame even with empty detections.
Returns the a similar array, where the last column is the object ID.
返回类似的数组，其中最后一列是对象ID。
NOTE: The number of objects returned may differ from the number of detections provided.
"""
self.frame_count += 1
#从现有的跟踪器获取预测的位置
#上一帧目标个数self.trackers
trks = np.zeros((len(self.trackers),5))#len(self.trackers)初始为0
print("trks: ", trks)
to_del = []
ret = []
for t,trk in enumerate(trks):
pos = self.trackers[t].predict()[0]#对于上一帧的目标，这一帧进行预测
print("pos: ", pos)
trk[:] = [pos[0], pos[1], pos[2], pos[3], 0]
#np.isnan()判断是否为空，np.any数组中只有有一个为true，则返回true
if(np.any(np.isnan(pos))):
to_del.append(t)#存放跟踪坐标数据为空的数据
#numpy.ma.masked_invaid屏蔽出现无效值的数组, numpy.ma.compress_rows压缩包含掩码值2-D数组的整行。
trks = np.ma.compress_rows(np.ma.masked_invalid(trks))
for t in reversed(to_del):#逆向删除异常的目标
self.trackers.pop(t)
#将检测结果指定给跟踪器
matched, unmatched_dets, unmatched_trks = associate_detections_to_trackers(dets,trks)
# 用指定的检测器，更新匹配到的跟踪器
for t, trk in enumerate(self.trackers):
#匈牙利算法没匹配到以及匹配到但是iou低于阈值的跟踪器
if(t not in unmatched_trks):#如果t是匹配到的目标
d = matched[np.where(matched[:,1]==t)[0],0]#匹配的跟踪器的编号与t相等，检测器的id
trk.update(dets[d,:][0])#利用检测器的结果更新卡尔曼
#对于没有匹配到的检测结果，初始化一个新的跟踪器
for i in unmatched_dets:
trk = KalmanBoxTracker(dets[i,:])
self.trackers.append(trk)
i = len(self.trackers)
for trk in reversed(self.trackers):
d = trk.get_state()[0]
#匹配到的时候hit_streak会加1
if((trk.time_since_update < 1) and (trk.hit_streak >= self.min_hits or self.frame_count <= self.min_hits)):
ret.append(np.concatenate((d,[trk.id+1])).reshape(1,-1)) # +1 as MOT benchmark requires positive
i -= 1
#remove dead tracklet
if(trk.time_since_update > self.max_age):
self.trackers.pop(i)
if(len(ret)>0):
return np.concatenate(ret)
return np.empty((0,5))
def parse_args():
"""Parse input arguments."""
parser = argparse.ArgumentParser(description='SORT demo')
parser.add_argument('--display', dest='display', help='Display online tracker output (slow) [False]',action='store_true')
args = parser.parse_args()
return args
if __name__ == '__main__':
#所有的视频序列
sequences = ['PETS09-S2L1','TUD-Campus','TUD-Stadtmitte','ETH-Bahnhof','ETH-Sunnyday','ETH-Pedcross2','KITTI-13','KITTI-17','ADL-Rundle-6','ADL-Rundle-8','Venice-2']
args = parse_args()#获取参数
display = args.display#是否显示跟踪过程
phase = 'train'
total_time = 0.0#总共耗时
total_frames = 0#视频帧总数
colours = np.random.rand(32,3) #32x3的随机矩阵，用于显示用
if(display):
#if not os.path.exists('mot_benchmark'):
#print('\n\tERROR: mot_benchmark link not found!\n\n Create a symbolic link to the MOT benchmark\n (https://motchallenge.net/data/2D_MOT_2015/#download). E.g.:\n\n $ ln -s /path/to/MOT2015_challenge/2DMOT2015 mot_benchmark\n\n')
#exit()
plt.ion()#打开交互模式
fig = plt.figure() #图片1
#如果没有output文件夹，则创建一个
if not os.path.exists('output'):
os.makedirs('output')
for seq in sequences:
mot_tracker = Sort() #创建一个跟踪器
print(seq)
#读取检测文本
seq_dets = np.loadtxt('data/%s/det/det.txt'%(seq),delimiter=',') #load detections
with open('output/%s.txt'%(seq),'w') as out_file:
print("Processing %s."%(seq))
#seq_dets[:,0].max()))视频帧数最大值
for frame in range(int(seq_dets[:,0].max())):
frame += 1 #detection and frame numbers begin at 1
dets = seq_dets[seq_dets[:,0]==frame,2:7]#每一帧的数据，第3到第6个数据
dets[:,2:4] += dets[:,0:2] #获取右下角的坐标=左上角坐标+宽和高
total_frames += 1
if(display):
ax1 = fig.add_subplot(111, aspect='equal')#1x1第一个子图
fn = 'data/%s/img1/%06d.jpg' % (seq, frame)#图像数据
#fn = 'mot_benchmark/%s/%s/img1/%06d.jpg'%(phase,seq,frame)
im =io.imread(fn)#读取每一帧图像
ax1.imshow(im)#显示图像
plt.title(seq+' Tracked Targets')
start_time = time.time()
#print(dets)
trackers = mot_tracker.update(dets)#利用检测的结果更新跟踪器,返回一个5个数的数组
#print(trackers)
cycle_time = time.time() - start_time
total_time += cycle_time
for d in trackers:
print('%d,%d,%.2f,%.2f,%.2f,%.2f,1,-1,-1,-1'%(frame,d[4],d[0],d[1],d[2]-d[0],d[3]-d[1]),file=out_file)
if(display):
d = d.astype(np.int32)#转换成整形
ax1.add_patch(patches.Rectangle((d[0],d[1]),d[2]-d[0],d[3]-d[1],fill=False,lw=3,ec=colours[d[4]%32,:]))
#ax1.set_adjustable('box-forced')
if(display):
fig.canvas.flush_events()
plt.draw()
ax1.cla()
print("Total Tracking took: %.3f for %d frames or %.1f FPS"%(total_time,total_frames,total_frames/total_time))
if(display):
print("Note: to get real runtime results run without the option: --display")
暂时未理解部分：
1.matched_indices = linear_assignment(-iou_matrix)
2.trks = np.ma.compress_rows(np.ma.masked_invalid(trks))
缺点：
1.这个算法直接利用已标注号的检测的结果，作为算法中的一部分。
2.待补充。
