我想将 cv2 Tracker 与连续的图像序列一起使用(以便执行实时跟踪)。因此,我不能使用vs = cv2.VideoCapture("video.mp4")
,我会有一堆numpy数组需要一个一个分析,有什么办法可以使用frame = vs.read()
来初始化跟踪器并更新跟踪器每个获取的图像?
我已经尝试获取一些图像,然后将它们转换为视频(mp4 或 avi),但是您可以想象这个过程非常缓慢,而且对实时跟踪毫无用处。
我想做这样的事情:
from tifffile import imread, imsave
img = imread("filename.tif") #shape = (1024,1024)
#HERE IS WHERE I HAVE PROBLEMS
vs = Treat_the_image_as_video(img) #In normal cases, here we would use cv2.VideoCapture
frame = vs.read() #?
bbox = get_bbox(image) #Get the bbox from the image (I already have this with cv2.findContours)
tracker = cv2.TrackerKCF_create() #Create tracker
tracker.init(frame, bbox)
while(True):
(success, bbox) = tracker.update(frame)
.
.
#Place the rectangle and text
.
.
img = imread("filename.tif") #shape = (1024,1024)
vs = Treat_the_image_as_video(img) #?
frame = vs.read() #?
如果您能就我如何解决此问题给我任何建议或提示,我将不胜感激。 提前致谢!!