宇宙仮面の研究室

Download YouTube

テスト用の動画をYouTubeから調達

import pafy
v = pafy.new("https://www.youtube.com/watch?v=xxxxxx")

print(v.title)
print(v.duration)
print(v.rating)
print(v.author)
print(v.length)
print(v.keywords)
print(v.thumb)
print(v.videoid)
print(v.viewcount)

streams = v.streams 
for i in streams: 
    print(i) 
      
<em># get best resolution regardless of format </em>
best = v.getbest() 
  
print(best.resolution, best.extension) 
  
<em># Download the video </em>
best.download()

import pafy

v = pafy.new("https://www.youtube.com/watch?v=xxxxxx")

print(v.title)

print(v.duration)

print(v.rating)

print(v.author)

print(v.length)

print(v.keywords)

print(v.thumb)

print(v.videoid)

print(v.viewcount)

streams = v.streams

for i in streams:

print(i)

# get best resolution regardless of format

best = v.getbest()

print(best.resolution, best.extension)

# Download the video

best.download()

PS C:\Users\仮面\Desktop\VideoTest> pip
Traceback (most recent call last):
File “d:\program files\python37\lib\runpy.py”, line 193, in run_module_as_main “main“, mod_spec) File “d:\program files\python37\lib\runpy.py”, line 85, in _run_code exec(code, run_globals) File “D:\Program Files\Python37\Scripts\pip.exe__main_.py”, line 5, in
ModuleNotFoundError: No module named ‘pip’

そこで、python -m ensurepip により、pip を修復

PS C:\Users\仮面\Desktop\VideoTest> python -m ensurepip
Looking in links: C:\Users\仮面\AppData\Local\Temp\tmp2qepvojk
Requirement already satisfied: setuptools in d:\program files\python37\lib\site-packages (40.8.0)
Collecting pip
Installing collected packages: pip
Successfully installed pip-19.0.3

Video の DCT, IDCT ローパス

OpenCvで、リアルタイムで、Video を200×200にスライスして、DCT→ローパス→IDCT を行ってみた。

左から、オリジナル、DCT、DCTの高周波成分をカット( *情報量はオリジナルの 1/100* )、IDCTで逆変換結果。

import numpy as np 
import cv2

cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)

while True:
    ret, frame = cap.read()
    if frame is None:
        print("frame is None")
        break

    frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    width = 200
    xstart = int((w-width)/2)
    xend = int(xstart + width)
    ystart = int((h-width)/2)
    yend = int(ystart + width)

    subFrame = frame[ystart:yend, xstart:xend]

    cv2.imshow('Frame', subFrame)
    dctFrame = cv2.dct(np.array(subFrame, dtype=np.float))
    
    cv2.imshow('DCT', dctFrame)
    
    f=20
    dctFrame[000:f, f:200]=0
    dctFrame[f:200, 000:f]=0
    dctFrame[f:200, f:200]=0
    
    idctFrame = cv2.idct(dctFrame)
    
    cv2.imshow('LP', dctFrame)
    cv2.imshow('IDCT', np.array(idctFrame, dtype=np.uint8))
    
    
    if cv2.waitKey(30) == 27:
        break

cap.release()
cv2.destroyAllWindows()

import numpy as np

import cv2

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

while True:

ret, frame = cap.read()

if frame is None:

print("frame is None")

break

frame = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

width = 200

xstart = int((w-width)/2)

xend = int(xstart + width)

ystart = int((h-width)/2)

yend = int(ystart + width)

subFrame = frame[ystart:yend, xstart:xend]

cv2.imshow('Frame', subFrame)

dctFrame = cv2.dct(np.array(subFrame, dtype=np.float))

cv2.imshow('DCT', dctFrame)

f=20

dctFrame[000:f, f:200]=0

dctFrame[f:200, 000:f]=0

dctFrame[f:200, f:200]=0

idctFrame = cv2.idct(dctFrame)

cv2.imshow('LP', dctFrame)

cv2.imshow('IDCT', np.array(idctFrame, dtype=np.uint8))

if cv2.waitKey(30) == 27:

break

cap.release()

cv2.destroyAllWindows()

OpenCVで、動き検出

【爆速】OpenCVで複数の物体検出　－ラズパイ－を参考に、物体認識を試してみた。

単一オブジェクトの動き検出

import numpy as np 
import cv2

ESC_KEY = 27

def get_bounding_box(X, Y):
    x_max = int(np.max(X))
    x_min = int(np.min(X))
    y_max = int(np.max(Y))
    y_min = int(np.min(Y))
    return (x_min, y_min, x_max, y_max)

def draw_bounding_box(img, bbox):
    img = cv2.rectangle(img, (bbox[0], bbox[1]), (bbox[2], bbox[3],), (0, 255, 0), 3)
    return img

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)

while True:  
    ret, frame = cap.read()
    if frame is None:
        break
    frame = cv2.GaussianBlur(frame, (7,7), 0)

    fgmask = fgbg.apply(frame)

    img1=frame
    img2=fgmask

    Y, X = np.where(fgmask &gt; 200)
    if len(X) == 0 or len(Y)==0 :
      continue

    bbox = get_bounding_box(X, Y)
    img2 = draw_bounding_box(img1, bbox)

    cv2.imshow('img1', img2)
    cv2.imshow('fgmask', fgmask)
    
    if cv2.waitKey(30) == ESC_KEY:
        break

cap.release()
cv2.destroyAllWindows()

import numpy as np

import cv2

ESC_KEY = 27

def get_bounding_box(X, Y):

x_max = int(np.max(X))

x_min = int(np.min(X))

y_max = int(np.max(Y))

y_min = int(np.min(Y))

return (x_min, y_min, x_max, y_max)

def draw_bounding_box(img, bbox):

img = cv2.rectangle(img, (bbox[0], bbox[1]), (bbox[2], bbox[3],), (0, 255, 0), 3)

return img

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

while True:

ret, frame = cap.read()

if frame is None:

break

frame = cv2.GaussianBlur(frame, (7,7), 0)

fgmask = fgbg.apply(frame)

img1=frame

img2=fgmask

Y, X = np.where(fgmask > 200)

if len(X) == 0 or len(Y)==0 :

continue

bbox = get_bounding_box(X, Y)

img2 = draw_bounding_box(img1, bbox)

cv2.imshow('img1', img2)

cv2.imshow('fgmask', fgmask)

if cv2.waitKey(30) == ESC_KEY:

break

cap.release()

cv2.destroyAllWindows()

KMeans を使って複数オブジェクトを認識をしてみたところ、ノイズがあると重心がうまく分離できず、Bounding Box が大きくずれてしまう。また、同一の物体であっても、複数のクラスターから構成されてしまうという問題点があった。

import numpy as np 
import cv2
from sklearn.cluster import KMeans

CLUSTER=10
colorlist = [(0,0,0),(0,0,255),(0,255,0),(0,255,255),(255,0,0),(255,0,255),(255,255,0),(255,255,255)]
ESC_KEY = 27

def get_bounding_box(X, Y, result, cluster):
    NO = np.where(result==cluster)
    x_max = int(np.max(X[NO]))
    x_min = int(np.min(X[NO]))
    y_max = int(np.max(Y[NO]))
    y_min = int(np.min(Y[NO]))
    return (x_min, y_min, x_max, y_max)

def draw_bounding_box(img, bbox):
    img = cv2.rectangle(img, (bbox[0], bbox[1]), (bbox[2], bbox[3],), (0, 255, 0), 3)
    return img

def draw_circle(img, X, Y, result, cluster):
    NO = np.where(result==cluster)
    for p in NO[0] :
      img = cv2.circle(img, (X[p], Y[p]), 5, colorlist[cluster%8],-1)
    return img

font = cv2.FONT_HERSHEY_SIMPLEX
def draw_number(img, X, Y, result, cluster):
    NO = np.where(result==cluster)
    for p in NO[0] :
      img = cv2.putText(img, str(cluster),(X[p], Y[p]), font, 1,colorlist[cluster%8],1,cv2.LINE_AA)
    return img

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture("video/vtest.avi") 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)

while True:  
    ret, frame = cap.read()
    if frame is None:
        break
    frame = cv2.GaussianBlur(frame, (5,5), 0)

    fgmask = fgbg.apply(frame)

    img1=frame
    img2=fgmask

    Y, X = np.where(fgmask &gt; 200)

    if X.size &lt; w :
      continue

    means = KMeans(n_clusters=CLUSTER, init='k-means++', n_jobs=1).fit_predict(np.array([X,Y]))

    for i in range(CLUSTER) :
      bbox = get_bounding_box(X, Y, means, i)
      
      img2 = draw_bounding_box(img1, bbox)
      img2 = draw_circle(img1, X, Y, means, i)
      <em># img2 = draw_number(img1, X, Y, means, i)</em>
      
    cv2.imshow('img1', img2)
    cv2.imshow('fgmask', fgmask)
    
    keyboard = cv2.waitKey(30)
    if keyboard == ESC_KEY:
        break

cap.release()
cv2.destroyAllWindows()

import numpy as np

import cv2

from sklearn.cluster import KMeans

CLUSTER=10

colorlist = [(0,0,0),(0,0,255),(0,255,0),(0,255,255),(255,0,0),(255,0,255),(255,255,0),(255,255,255)]

ESC_KEY = 27

def get_bounding_box(X, Y, result, cluster):

NO = np.where(result==cluster)

x_max = int(np.max(X[NO]))

x_min = int(np.min(X[NO]))

y_max = int(np.max(Y[NO]))

y_min = int(np.min(Y[NO]))

return (x_min, y_min, x_max, y_max)

def draw_bounding_box(img, bbox):

img = cv2.rectangle(img, (bbox[0], bbox[1]), (bbox[2], bbox[3],), (0, 255, 0), 3)

return img

def draw_circle(img, X, Y, result, cluster):

NO = np.where(result==cluster)

for p in NO[0] :

img = cv2.circle(img, (X[p], Y[p]), 5, colorlist[cluster%8],-1)

return img

font = cv2.FONT_HERSHEY_SIMPLEX

def draw_number(img, X, Y, result, cluster):

NO = np.where(result==cluster)

for p in NO[0] :

img = cv2.putText(img, str(cluster),(X[p], Y[p]), font, 1,colorlist[cluster%8],1,cv2.LINE_AA)

return img

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture("video/vtest.avi")

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

while True:

ret, frame = cap.read()

if frame is None:

break

frame = cv2.GaussianBlur(frame, (5,5), 0)

fgmask = fgbg.apply(frame)

img1=frame

img2=fgmask

Y, X = np.where(fgmask > 200)

if X.size < w :

continue

means = KMeans(n_clusters=CLUSTER, init='k-means++', n_jobs=1).fit_predict(np.array([X,Y]))

for i in range(CLUSTER) :

bbox = get_bounding_box(X, Y, means, i)

img2 = draw_bounding_box(img1, bbox)

img2 = draw_circle(img1, X, Y, means, i)

# img2 = draw_number(img1, X, Y, means, i)

cv2.imshow('img1', img2)

cv2.imshow('fgmask', fgmask)

keyboard = cv2.waitKey(30)

if keyboard == ESC_KEY:

break

cap.release()

cv2.destroyAllWindows()

次に、オブジェクト輪郭検出を参考に複数のオブジェクトの動き認識してみた。動いていないと認識できない、物体の特定ができない、といった問題があり、正確ではないけれど、ある程度認識することができる。

import numpy as np 
import cv2

ESC_KEY = 27

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture("video/vtest.avi") 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)

while True:  
  ret, frame = cap.read()
  if frame is None:
      break
  frame = cv2.GaussianBlur(frame, (7,7), 0)

  fgmask = fgbg.apply(frame)

  # find contours
  labels, contours, hierarchy = cv2.findContours(fgmask, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)

  for i in range(0, len(contours)):
      if len(contours[i]) > 0:

          # remove small objects
          if cv2.contourArea(contours[i]) &lt; 500:
              continue

          # Draw Bounding Box
          rect = contours[i]
          x, y, w, h = cv2.boundingRect(rect)
          cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 3)

          # Draw Contours
          # frame = cv2.drawContours(frame, contours, i, (0,255,0), 3)

  cv2.imshow('frame', frame)
  cv2.imshow('fgmask', fgmask)
  
  keyboard = cv2.waitKey(int(1000/fps))
  if keyboard == ESC_KEY:
      break

cap.release()
cv2.destroyAllWindows()

import numpy as np

import cv2

ESC_KEY = 27

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture("video/vtest.avi")

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

while True:

ret, frame = cap.read()

if frame is None:

break

frame = cv2.GaussianBlur(frame, (7,7), 0)

fgmask = fgbg.apply(frame)

# find contours

labels, contours, hierarchy = cv2.findContours(fgmask, cv2.RETR_LIST, cv2.CHAIN_APPROX_NONE)

for i in range(0, len(contours)):

if len(contours[i]) > 0:

# remove small objects

if cv2.contourArea(contours[i]) < 500:

continue

# Draw Bounding Box

rect = contours[i]

x, y, w, h = cv2.boundingRect(rect)

cv2.rectangle(frame, (x, y), (x + w, y + h), (0, 255, 0), 3)

# Draw Contours

# frame = cv2.drawContours(frame, contours, i, (0,255,0), 3)

cv2.imshow('frame', frame)

cv2.imshow('fgmask', fgmask)

keyboard = cv2.waitKey(int(1000/fps))

if keyboard == ESC_KEY:

break

cap.release()

cv2.destroyAllWindows()

drawContours だけで輪郭を描画できる。

OpenCv で動きを検出したらビデオに出力

動きを検出したら、ビデオき書き出す

import numpy as np 
import cv2

ESC_KEY = 27
THRESHOLD = 100000

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)

fourcc = cv2.VideoWriter_fourcc(*'MP4V') 
out = cv2.VideoWriter('out2.mp4',fourcc, 30.0, (int(w),int(h))) 

while True:
    ret, frame = cap.read()
    if frame is None:
        break

    fgmask = fgbg.apply(frame)
    cv2.imshow('frame', frame)
    cv2.imshow('fgmask', fgmask)
    
    motion = sum(sum(fgmask))
    if motion &gt; THRESHOLD:
        print("MOTION DETECTED:",str(motion))
        out.write(frame)

    keyboard = cv2.waitKey(30)
    if keyboard == ESC_KEY:
        break

cap.release()
out.release() 
cv2.destroyAllWindows()

import numpy as np

import cv2

ESC_KEY = 27

THRESHOLD = 100000

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

fourcc = cv2.VideoWriter_fourcc(*'MP4V')

out = cv2.VideoWriter('out2.mp4',fourcc, 30.0, (int(w),int(h)))

while True:

ret, frame = cap.read()

if frame is None:

break

fgmask = fgbg.apply(frame)

cv2.imshow('frame', frame)

cv2.imshow('fgmask', fgmask)

motion = sum(sum(fgmask))

if motion > THRESHOLD:

print("MOTION DETECTED:",str(motion))

out.write(frame)

keyboard = cv2.waitKey(30)

if keyboard == ESC_KEY:

break

cap.release()

out.release()

cv2.destroyAllWindows()

OpenCvで timelapse

既存のビデオファイルから、タイムラプスを作成する方法

import cv2
import numpy as np

RATE = 10 <em># 10倍速</em>
cap = cv2.VideoCapture('AB.mp4') 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)
fourcc = cv2.VideoWriter_fourcc(*'MP4V') 
out = cv2.VideoWriter('out.mp4',fourcc, fps, (int(w),int(h))) 

count=RATE
while(True):
    ret, frame = cap.read()
    if ret==True:
        if count==0:
            out.write(frame)
            count=RATE
        count-=1
        cv2.imshow('frame',frame)  
        keyboard = cv2.waitKey(1)
        if keyboard == 27:
            break
    else:
        break
cap.release()
out.release()  
cv2.destroyAllWindows()

import cv2

import numpy as np

RATE = 10 # 10倍速

cap = cv2.VideoCapture('AB.mp4')

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

fourcc = cv2.VideoWriter_fourcc(*'MP4V')

out = cv2.VideoWriter('out.mp4',fourcc, fps, (int(w),int(h)))

count=RATE

while(True):

ret, frame = cap.read()

if ret==True:

if count==0:

out.write(frame)

count=RATE

count-=1

cv2.imshow('frame',frame)

keyboard = cv2.waitKey(1)

if keyboard == 27:

break

else:

break

cap.release()

out.release()

cv2.destroyAllWindows()

fourcc ：　Windows 10 の場合、次の動作を確認した。

フォーマット	fourcc
avi	DIVX
avi	DIVX
wmv	WMV1
wmv	WNV2
mp4	MP4V

OpenCV でキャプチャーしたビデオをファイルに出力

参考
https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_gui/py_video_display/py_video_display.html

import cv2

cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)
fourcc = cv2.VideoWriter_fourcc(*'MP4V') 
out = cv2.VideoWriter('out.mp4',fourcc, 30.0, (int(w),int(h))) 

while(True):
    ret, frame = cap.read()
    if ret==True:
        out.write(frame)       
        cv2.imshow('frame',frame)  
        keyboard = cv2.waitKey(30)
        if keyboard == 27:
            break
    else:
        break
cap.release()
out.release()  
cv2.destroyAllWindows()

import cv2

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

fourcc = cv2.VideoWriter_fourcc(*'MP4V')

out = cv2.VideoWriter('out.mp4',fourcc, 30.0, (int(w),int(h)))

while(True):

ret, frame = cap.read()

if ret==True:

out.write(frame)

cv2.imshow('frame',frame)

keyboard = cv2.waitKey(30)

if keyboard == 27:

break

else:

break

cap.release()

out.release()

cv2.destroyAllWindows()

OpenCV で背景差分

フレームを直接操作した場合

import cv2

ESC_KEY = 27
cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)
avg = None

while True:
    ret, frame = cap.read()
    if frame is None:
        break
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    if avg is None:
        avg = gray.copy().astype("float")
        continue
    cv2.accumulateWeighted(gray, avg, 0.5)
    frameDelta = cv2.absdiff(gray, cv2.convertScaleAbs(avg))
    ret, thresh = cv2.threshold(frameDelta, 20, 255, cv2.THRESH_BINARY)

    cv2.imshow('frame', frame)
    cv2.imshow('thresh', thresh)

    keyboard = cv2.waitKey(30)
    if keyboard == ESC_KEY:
        break

cap.release()
cv2.destroyAllWindows()

import cv2

ESC_KEY = 27

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

avg = None

while True:

ret, frame = cap.read()

if frame is None:

break

gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

if avg is None:

avg = gray.copy().astype("float")

continue

cv2.accumulateWeighted(gray, avg, 0.5)

frameDelta = cv2.absdiff(gray, cv2.convertScaleAbs(avg))

ret, thresh = cv2.threshold(frameDelta, 20, 255, cv2.THRESH_BINARY)

cv2.imshow('frame', frame)

cv2.imshow('thresh', thresh)

keyboard = cv2.waitKey(30)

if keyboard == ESC_KEY:

break

cap.release()

cv2.destroyAllWindows()

Numpy で書いてみた

import numpy as np 
import cv2

THRESHOLD = 100000
cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)
if fps == 0:
    fps=30.0
  
old64 = np.zeros((int(h),int(w)),  dtype=np.int64)

while True:
    ret, frame = cap.read()
    if frame is None:
        break
    gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)
    gray64 = np.array(gray, np.int64)
    if old64 is not None:
        diff64 = abs(gray64-old64)
        old64 = gray64
    cv2.imshow('npframe', gray)
    res64=np.where(diff64 &gt; 50, 255, 0)
    motion = sum(sum(res64))
    if motion &gt; THRESHOLD:
        print("MOTION DETECTED:",str(motion))
    cv2.imshow('Frame', np.array(res64, np.uint8))

    keyboard = cv2.waitKey(30)
    if keyboard == ESC_KEY:
        break

cap.release()
out.release() 
cv2.destroyAllWindows()

import numpy as np

import cv2

THRESHOLD = 100000

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

if fps == 0:

fps=30.0

old64 = np.zeros((int(h),int(w)), dtype=np.int64)

while True:

ret, frame = cap.read()

if frame is None:

break

gray = cv2.cvtColor(frame, cv2.COLOR_BGR2GRAY)

gray64 = np.array(gray, np.int64)

if old64 is not None:

diff64 = abs(gray64-old64)

old64 = gray64

cv2.imshow('npframe', gray)

res64=np.where(diff64 > 50, 255, 0)

motion = sum(sum(res64))

if motion > THRESHOLD:

print("MOTION DETECTED:",str(motion))

cv2.imshow('Frame', np.array(res64, np.uint8))

keyboard = cv2.waitKey(30)

if keyboard == ESC_KEY:

break

cap.release()

out.release()

cv2.destroyAllWindows()

BackgroundSubtractorMOG でもっと簡単に書ける

背景差分
bgsegm は、Background-Foreground Segmentation Methods の略

import numpy as np
import cv2

cap = cv2.VideoCapture('out.mp4')

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

while(1):
    ret, frame = cap.read()

    fgmask = fgbg.apply(frame)

    cv2.imshow('frame',frame)
    cv2.imshow('fgmask',fgmask)
    k = cv2.waitKey(30)
    if k == 27:
        break

cap.release()
cv2.destroyAllWindows()

import numpy as np

import cv2

cap = cv2.VideoCapture('out.mp4')

fgbg = cv2.bgsegm.createBackgroundSubtractorMOG()

while(1):

ret, frame = cap.read()

fgmask = fgbg.apply(frame)

cv2.imshow('frame',frame)

cv2.imshow('fgmask',fgmask)

k = cv2.waitKey(30)

if k == 27:

break

cap.release()

cv2.destroyAllWindows()

numpy で2次元配列化からスライス

import numpy as np 

a=np.arange(8*8).reshape((8,8))
print(a)

print(a[0:4, 0:4])
print(a[4:8, 0:4])
print(a[0:4, 4:8])
print(a[4:8, 4:8])

import numpy as np

a=np.arange(8*8).reshape((8,8))

print(a)

print(a[0:4, 0:4])

print(a[4:8, 0:4])

print(a[0:4, 4:8])

print(a[4:8, 4:8])

実行結果
オリジナル配列
[[ 0 1 2 3 4 5 6 7]
[ 8 9 10 11 12 13 14 15]
[16 17 18 19 20 21 22 23]
[24 25 26 27 28 29 30 31]
[32 33 34 35 36 37 38 39]
[40 41 42 43 44 45 46 47]
[48 49 50 51 52 53 54 55]
[56 57 58 59 60 61 62 63]]

print(a[0:4, 0:4])
[[ 0 1 2 3]
[ 8 9 10 11]
[16 17 18 19]
[24 25 26 27]]

print(a[4:8, 0:4])
[[32 33 34 35]
[40 41 42 43]
[48 49 50 51]
[56 57 58 59]]

print(a[0:4, 4:8])
[[ 4 5 6 7]
[12 13 14 15]
[20 21 22 23]
[28 29 30 31]]

print(a[4:8, 4:8])
[[36 37 38 39]
[44 45 46 47]
[52 53 54 55]
[60 61 62 63]]

ビデオから、部分画像をスライス

import numpy as np 
import cv2

cap = cv2.VideoCapture(0) 
w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)
h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)
fps = cap.get(cv2.CAP_PROP_FPS)

while True:
    ret, frame = cap.read()
    if frame is None:
        print("frame is None")
        break

    width = 200
    xstart = int((w-width)/2)
    xend = int(xstart + width)
    ystart = int((h-width)/2)
    yend = int(ystart + width)

    subFrame = frame[ystart:yend, xstart:xend]
    cv2.imshow('Frame', frame)
    cv2.imshow('SubFrame', subFrame)
    
    if cv2.waitKey(30) == 27:
        break

cap.release()
cv2.destroyAllWindows()

import numpy as np

import cv2

cap = cv2.VideoCapture(0)

w = cap.get(cv2.CAP_PROP_FRAME_WIDTH)

h = cap.get(cv2.CAP_PROP_FRAME_HEIGHT)

fps = cap.get(cv2.CAP_PROP_FPS)

while True:

ret, frame = cap.read()

if frame is None:

print("frame is None")

break

width = 200

xstart = int((w-width)/2)

xend = int(xstart + width)

ystart = int((h-width)/2)

yend = int(ystart + width)

subFrame = frame[ystart:yend, xstart:xend]

cv2.imshow('Frame', frame)

cv2.imshow('SubFrame', subFrame)

if cv2.waitKey(30) == 27:

break

cap.release()

cv2.destroyAllWindows()

opencv Tutorial 顔認識

opencv Tutorial Cascade Classifier

import cv2 as cv
def detectAndDisplay(frame):
    frame_gray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)
    frame_gray = cv.equalizeHist(frame_gray)
    <em>#-- Detect faces</em>
    faces = face_cascade.detectMultiScale(frame_gray)
    for (x,y,w,h) in faces:
        center = (x + w//2, y + h//2)
        frame = cv.ellipse(frame, center, (w//2, h//2), 0, 0, 360, (255, 0, 255), 4)
    cv.imshow('Capture - Face detection', frame)
face_cascade = cv.CascadeClassifier()
if not face_cascade.load('Haar-cascade/haarcascade_frontalface_alt.xml'):
    print('--(!)Error loading face cascade')
    exit(0)
cap = cv.VideoCapture(0)
if not cap.isOpened:
    print('--(!)Error opening video capture')
    exit(0)
while True:
    ret, frame = cap.read()
    if frame is None:
        print('--(!) No captured frame -- Break!')
        break
    detectAndDisplay(frame)
    if cv.waitKey(10) == 27:
        break

import cv2 as cv

def detectAndDisplay(frame):

frame_gray = cv.cvtColor(frame, cv.COLOR_BGR2GRAY)

frame_gray = cv.equalizeHist(frame_gray)

#-- Detect faces

faces = face_cascade.detectMultiScale(frame_gray)

for (x,y,w,h) in faces:

center = (x + w//2, y + h//2)

frame = cv.ellipse(frame, center, (w//2, h//2), 0, 0, 360, (255, 0, 255), 4)

cv.imshow('Capture - Face detection', frame)

face_cascade = cv.CascadeClassifier()

if not face_cascade.load('Haar-cascade/haarcascade_frontalface_alt.xml'):

print('--(!)Error loading face cascade')

exit(0)

cap = cv.VideoCapture(0)

if not cap.isOpened:

print('--(!)Error opening video capture')

exit(0)

while True:

ret, frame = cap.read()

if frame is None:

print('--(!) No captured frame -- Break!')

break

detectAndDisplay(frame)

if cv.waitKey(10) == 27:

break