how to test model on my sample video

[alighofrani95]

Hi dears,
sorry for send email directly,

I'm interested in I3D method for action recognition and i cloned the github project "https://github.com/deepmind/kinetics-i3d"
i want to test the model with my sample recorded video but don't know how to convert my *.mp4 video file (25fps rgb) to *rgb.npz and *flow.npz

could it possible to help or share me a reprocessing script file

best regards.

[heinrich-heine]

Hi, I wrote a simple script to apply on gif files:

#open the cricket shot video under data/ and convert it into a npy array of shape (1, num_frames, 224, 224, 3)

from PIL import Image
from PIL import GifImagePlugin
import numpy as np
import os
import re
import shutil

#delete the folder test_files
shutil.rmtree('test_files', ignore_errors=True)
#create new empty one
os.mkdir('test_files')

#select the .gif
imageObject = Image.open('data/dog.gif')

print(imageObject.is_animated)

print(imageObject.n_frames)

Display individual frames from the loaded animated GIF file

for frame in range(0,imageObject.n_frames):

imageObject.seek(frame)
image_name = frame
resized_im = imageObject.resize((224,224))

resized_im.save('test_files/' + str(frame) + '.png')

#get the filenames with os
file_list = os.listdir('test_files/')
#sort maybe that matters

#to not take ds store

#ONLY needed when .DS store file in the dir, if craeted newly by script--> no .DS
#file_list = file_list[1:]

file_list.sort(key=lambda f: int(re.sub('\D', '', f)))

#pipeline which can be easily adapted for my data in image files

#base of stack
pic = Image.open('test_files/0.png')
pic = pic.convert('RGB')
pix = np.array(pic.getdata()).reshape(pic.size[0], pic.size[1], 3)
video_array = np.reshape(pix, (1, 224, 224,3))

for file in range(1,len(file_list)):

pic = Image.open('test_files/' + file_list[file])
pic = pic.convert('RGB')
pix = np.array(pic.getdata()).reshape(pic.size[0], pic.size[1], 3)
pix = np.reshape(pix, (1, 224, 224,3))

#stack all frame arrays or better concatenate
video_array = np.concatenate((video_array, pix))

def convert_pixel_value(pixel_value):

converted = ((pixel_value-128)/128) * 1
    
return converted

video_array = convert_pixel_value(video_array)

frames = (np.shape(video_array))[0]

np.save('data/input_array.npy', video_array)

video_array = np.reshape(video_array, (1, frames, 224, 224, 3))