ai-content-maker/.venv/Lib/site-packages/moviepy/video/tools/tracking.py

187 lines
5.9 KiB
Python
Raw Permalink Normal View History

2024-05-11 23:00:43 +03:00
"""
This module contains different functions for tracking objects in videos,
manually or automatically. The tracking functions return results under
the form: ``( txy, (fx,fy) )`` where txy is of the form [(ti, xi, yi)...]
and (fx(t),fy(t)) give the position of the track for all times t (if the
time t is out of the time bounds of the tracking time interval
fx and fy return the position of the object at the start or at the end
of the tracking time interval).
"""
import numpy as np
from moviepy.decorators import convert_to_seconds, use_clip_fps_by_default
from ..io.preview import imdisplay
from .interpolators import Trajectory
try:
import cv2
autotracking_possible = True
except:
# Note: this will be later fixed with scipy/skimage replacements
# but for the moment OpenCV is mandatory, so...
autotracking_possible = False
# WE START WITH A TOOL FUNCTION
# MANUAL TRACKING
@convert_to_seconds(["t1","t2"])
@use_clip_fps_by_default
def manual_tracking(clip, t1=None, t2=None, fps=None, nobjects = 1,
savefile = None):
"""
Allows manual tracking of an object(s) in the video clip between
times `t1` and `t2`. This displays the clip frame by frame
and you must click on the object(s) in each frame. If ``t2=None``
only the frame at ``t1`` is taken into account.
Returns a list [(t1,x1,y1),(t2,x2,y2) etc... ] if there is one
object per frame, else returns a list whose elements are of the
form (ti, [(xi1,yi1), (xi2,yi2), ...] )
Parameters
-------------
t1,t2:
times during which to track (defaults are start and
end of the clip). t1 and t2 can be expressed in seconds
like 15.35, in (min, sec), in (hour, min, sec), or as a
string: '01:03:05.35'.
fps:
Number of frames per second to freeze on. If None, the clip's
fps attribute is used instead.
nobjects:
Number of objects to click on each frame.
savefile:
If provided, the result is saved to a file, which makes
it easier to edit and re-use later.
Examples
---------
>>> from moviepy.editor import VideoFileClip
>>> from moviepy.video.tools.tracking import manual_tracking
>>> clip = VideoFileClip("myvideo.mp4")
>>> # manually indicate 3 trajectories, save them to a file
>>> trajectories = manual_tracking(clip, t1=5, t2=7, fps=5,
nobjects=3, savefile="track.txt")
>>> # ...
>>> # LATER, IN ANOTHER SCRIPT, RECOVER THESE TRAJECTORIES
>>> from moviepy.video.tools.tracking import Trajectory
>>> traj1, traj2, traj3 = Trajectory.load_list('track.txt')
>>> # If ever you only have one object being tracked, recover it with
>>> traj, = Trajectory.load_list('track.txt')
"""
import pygame as pg
screen = pg.display.set_mode(clip.size)
step = 1.0 / fps
if (t1 is None) and (t2 is None):
t1,t2 = 0, clip.duration
elif (t2 is None):
t2 = t1 + step / 2
t = t1
txy_list = []
def gatherClicks(t):
imdisplay(clip.get_frame(t), screen)
objects_to_click = nobjects
clicks = []
while objects_to_click:
for event in pg.event.get():
if event.type == pg.KEYDOWN:
if (event.key == pg.K_BACKSLASH):
return "return"
elif (event.key == pg.K_ESCAPE):
raise KeyboardInterrupt()
elif event.type == pg.MOUSEBUTTONDOWN:
x, y = pg.mouse.get_pos()
clicks.append((x, y))
objects_to_click -= 1
return clicks
while t < t2:
clicks =gatherClicks(t)
if clicks == 'return':
txy_list.pop()
t -= step
else:
txy_list.append((t,clicks))
t += step
tt, xylist = zip(*txy_list)
result = []
for i in range(nobjects):
xys = [e[i] for e in xylist]
xx, yy = zip(*xys)
result.append(Trajectory(tt, xx, yy))
if savefile is not None:
Trajectory.save_list(result, savefile)
return result
# AUTOMATED TRACKING OF A PATTERN
def findAround(pic,pat,xy=None,r=None):
"""
find image pattern ``pat`` in ``pic[x +/- r, y +/- r]``.
if xy is none, consider the whole picture.
"""
if xy and r:
h,w = pat.shape[:2]
x,y = xy
pic = pic[y-r : y+h+r , x-r : x+w+r]
matches = cv2.matchTemplate(pat,pic,cv2.TM_CCOEFF_NORMED)
yf,xf = np.unravel_index(matches.argmax(),matches.shape)
return (x-r+xf,y-r+yf) if (xy and r) else (xf,yf)
def autoTrack(clip, pattern, tt=None, fps=None, radius=20, xy0=None):
"""
Tracks a given pattern (small image array) in a video clip.
Returns [(x1,y1),(x2,y2)...] where xi,yi are
the coordinates of the pattern in the clip on frame i.
To select the frames you can either specify a list of times with ``tt``
or select a frame rate with ``fps``.
This algorithm assumes that the pattern's aspect does not vary much
and that the distance between two occurences of the pattern in
two consecutive frames is smaller than ``radius`` (if you set ``radius``
to -1 the pattern will be searched in the whole screen at each frame).
You can also provide the original position of the pattern with xy0.
"""
if not autotracking_possible:
raise IOError("Sorry, autotrack requires OpenCV for the moment. "
"Install OpenCV (aka cv2) to use it.")
if not xy0:
xy0 = findAround(clip.get_frame(tt[0]),pattern)
if tt is None:
tt = np.arange(0, clip.duration, 1.0/fps)
xys = [xy0]
for t in tt[1:]:
xys.append( findAround(clip.get_frame(t),pattern,
xy=xys[-1],r=radius))
xx,yy = zip(*xys)
return Trajectory(tt, xx, yy)