Python + Hachoir-Metadata - Reading FPS tag from .MP4 file

Question

I'm writing a Windows application in Python which has to read metadata from an .MP4 video file.

I started writing the application in Python 3, but was unable to find a suitable module to read metadata from video files. That's when I used 3to2 to move the entire project to Python 2, so I could install Hachoir-Metadata, which was praised all over the net, using pip install hachoir-core, pip install hachoir-parser, and pip install hachoir-metadata

I used the following code:

from hachoir_core.error import HachoirError
from hachoir_core.cmd_line import unicodeFilename
from hachoir_parser import createParser
from hachoir_core.tools import makePrintable
from hachoir_metadata import extractMetadata
from hachoir_core.i18n import getTerminalCharset

# Get metadata for video file
def metadata_for(filename):

    filename, realname = unicodeFilename(filename), filename
    parser = createParser(filename, realname)
    if not parser:
        print "Unable to parse file"
        exit(1)
    try:
        metadata = extractMetadata(parser)
    except HachoirError, err:
        print "Metadata extraction error: %s" % unicode(err)
        metadata = None
    if not metadata:
        print "Unable to extract metadata"
        exit(1)

    text = metadata.exportPlaintext()
    charset = getTerminalCharset()
    for line in text:
        print makePrintable(line, charset)

    return metadata

pathname = c:/video.mp4
meta = metadata_for(pathname)
print meta

This returned the following metadata:

Duration: 37 sec 940 ms
Image width: 1280 pixels
Image height: 960 pixels
Creation date: 2014-12-13 19:27:36
Last modification: 2014-12-13 19:27:36
Comment: Play speed: 100.0%
Comment: User volume: 100.0%
MIME type: video/quicktime
Endianness: Big endian

This is great, except for the fact that I also really need to know the frames per second (FPS).. For .AVI files Hachoir-Metadata does show the FPS, as you can see from this test output:

Duration: 6 sec 66 ms
Image width: 256 pixels
Image height: 240 pixels
Frame rate: 30.0 fps
Bit rate: 884.4 Kbit/sec
Comment: Has audio/video index (2920 bytes)
MIME type: video/x-msvideo
Endianness: Little endian

And yes, the FPS tag is set on the .MP4 file (100fps).

Is there a way to extract the FPS from a .MP4 file? Preferably including width(px), height(px), duration, and creation time as well.

Thanks in advance for any help!

The data that you have shown dose include all the information that you requested __but__ 100 fps sounds _unlikely_ for a video file - MP4/quicktime does not truly have a frame rate as it uses keyframes and interpolation frames I suspect that whatever is reporting `100 fps` is incorrectly labeling the `Play Speed=100%` and Hachoir is correct. — Steve Barnes, Jan 09 '15 at 06:08
@SteveBarnes Thanks for your quick answer! The video was shot using a GoPro HERO3 Black edition, which is capable of shooting 100 fps. Windows explorer > file properties > details tab also shows the 100 fps. — kregus, Jan 09 '15 at 08:42

score 4 · Accepted Answer · edited May 23 '17 at 11:53

Ok, I managed to extract all the data I need and more! This answer on Stack Overflow gave me the idea to try MediaInfo to extract metadata.

For this I switched back to Python 3 again. I also had to change line 22 in MediaInfoDLL3.py to MediaInfoDLL_Handler = WinDLL("C:\Program Files (x86)\MediaInfo\MediaInfo_i386.dll")

This is the code I used:

import os

os.chdir(os.environ["PROGRAMFILES"] + "\\mediainfo")  # The folder where you installed MediaInfo
from MediaInfoDLL3 import MediaInfo, Stream

MI = MediaInfo()

def get_mediainfo_from(directory):
  for file in os.listdir(directory):
    MI.Open(directory + file)
    file_extension = MI.Get(Stream.General, 0, "FileExtension")
    duration_string = MI.Get(Stream.Video, 0, "Duration/String3")  # Length. "Duration" for ms
    fps_string = MI.Get(Stream.Video, 0, "FrameRate")
    width_string = MI.Get(Stream.Video, 0, "Width")
    height_string = MI.Get(Stream.Video, 0, "Height")
    aspect_ratio_string = MI.Get(Stream.Video, 0, "DisplayAspectRatio")
    frames_string = MI.Get(Stream.Video, 0, "FrameCount")
    local_created_date_string = MI.Get(Stream.General, 0, "File_Created_Date_Local")  # Date of copying
    local_modified_date_string = MI.Get(Stream.General, 0, "File_Modified_Date_Local")  # Date of filming

    if file_extension == "MP4":
      print("Extension: "+file_extension)
      print("Length: "+duration_string)
      print("FPS: "+fps_string)
      print("Width: "+width_string)
      print("Height: "+height_string)
      print("Ratio: "+aspect_ratio_string)
      print("Frames: "+frames_string)
      print("Created Date: "+local_created_date_string)
      print("Modified Date: "+local_modified_date_string)

    else:
      print("{} ain't no MP4 file!".format(file))

    MI.Close()

get_mediainfo_from("C:\\Users\\Nick\\Desktop\\test\\")  # The folder with video files

# print(MI.Option("Info_Parameters"))  # Show list of available metadata tags

This returned:

Extension: MP4
Length: 00:00:37.940
FPS: 100.000
Width: 1280
Height: 960
Ratio: 1.333
Frames: 3794
Created Date: 2015-01-07 15:25:11.678
Modified Date: 2014-12-13 19:28:14.000

Hopefully this helps someone!

Thanks for sharing, this is helpful! – emf Sep 03 '16 at 12:53 — emf, Sep 03 '16 at 12:53

score 1 · Answer 2 · answered Nov 10 '16 at 11:23

This is totally different from the way you have approached with the problem. I wanted to get only the fps of a mp4 video and I got it by using openCV. This is not an answer but I thought it will be useful for someone.

import cv2
cap = cv2.VideoCapture('name_of_video_file')
fps    = cap.get(cv2.CAP_PROP_FPS)
print 'fps=',fps

You can get other metadata also in the same way. For eg.,

length_of_video = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
width  = int(cap.get(cv2.CAP_PROP_FRAME_WIDTH))
height = int(cap.get(cv2.CAP_PROP_FRAME_HEIGHT))

This site will help you with the keywords: http://docs.opencv.org/3.1.0/dd/de7/group__videoio.html

Python + Hachoir-Metadata - Reading FPS tag from .MP4 file

2 Answers2