autonomous_video_cropper.py

# -*- coding: utf-8 -*-
"""Autonomous Video Cropper.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1qBQFFcG7o06MaUefVgrVE6m5coCcll0V
"""

#!pip install progressbar
#!pip install pytube
#!pip install cvlib

import cvlib as cv
from cvlib.object_detection import draw_bbox
import cv2

import cvlib as cv
from cvlib.object_detection import draw_bbox
import cv2

from pytube import YouTube
vid_url = 'https://www.youtube.com/watch?v=dsOZpD1j5Ac'
YouTube(vid_url).streams.filter(progressive=True, file_extension='mp4').first().download()

from glob import glob
file = glob('*.mp4')
og_vid = file[0]
og_vid

cap = cv2.VideoCapture(og_vid)
framecount = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter('output.mp4v',fourcc, 30, (800,880), True)

frame = 0

boxes = []
labels = []
confs = []
frames = []
print('Processing: %s' %og_vid)
while frame < framecount:
  ret , image = cap.read()
  frames.append(image)
  frame += 1
  if (type(image) == type(None)):
    break
  # apply object detection
  bbox, label, conf = cv.detect_common_objects(image)
  boxes.append(bbox)
  labels.append(label)
  confs.append(conf)
  #print(bbox, label, conf)
  print('Frame: {} / {}'.format(frame,framecount))
  
  # draw bounding box over detected objects
  box = draw_bbox(image, bbox, label, conf)
  out.write(box)

  # press any key to close window           
  if cv2.waitKey(1) == ord('x'):
        break
# release resources
cap.release()

len(frames)

def pad_img_to_fit_box(img, x1, x2, y1, y2):
  img = cv2.copyMakeBorder(img, - min(0, y1), max(y2 - img.shape[0], 0),
                            -min(0, x1), max(x2 - img.shape[1], 0),cv2.BORDER_REPLICATE)
  y2 += -min(0, y1)
  y1 += -min(0, y1)
  x2 += -min(0, x1)
  x1 += -min(0, x1)
  return img, x1, x2, y1, y2

def imcrop(img, box):
  x1, y1, x2, y2 = box
  if x1 < 0 or y1 < 0 or x2 > img.shape[1] or y2 > img.shape[0]:
    img, x1, x2, y1, y2 = pad_img_to_fit_box(img, x1, x2, y1, y2)
  return img[y1:y2, x1:x2, :]

print('Filtering out objects that are not people')
indx = [i.index('person') for i in labels]
pbox = []
for i in range(0,framecount):
  pbox.append(boxes[i][indx[i]])

video_name = 'cropped.mp4v'

nframe = frames[0]

video = cv2.VideoWriter(video_name,fourcc, 20.0, (800,800))

print('Cropping Video...')
for im in range(len(frames)):
    video.write(imcrop(frames[im],pbox[im]))
video.release()
print('Completed\nCropped Video Located: %s' % video_name)

import matplotlib.pyplot as plt

plt.imshow(imcrop(frames[50],pbox[50]))