Module torchvid
Class ImageFrame
| ImageFrame:to_byte_tensor () | Copies video frame pixel data into a torch.ByteTensor. |
| ImageFrame:to_float_tensor () | Copies video frame pixel data into a torch.FloatTensor. |
| ImageFrame:timestamp () | Get the timestamp of this image frame (in seconds). |
Class Video
| Video.new (path) | Creates a new Video. |
| Video:duration () | Get the duration of the video in seconds. |
| Video:guess_image_frame_rate () | Guess the average frame rate of the video in FPS. |
| Video:get_image_frame_count () | Get the number of image frames in the video. |
| Video:filter (pixel_format_name[, filterchain='null']) | Apply a filterchain to the video. |
| Video:next_image_frame () | Read the next video frame from the video. |
| Video:seek (seek_target) | Seek to the first keyframe before the frame number specified. |
Class ImageFrame
- ImageFrame:to_byte_tensor ()
-
Copies video frame pixel data into a
torch.ByteTensor.Returns:
-
torch.ByteTensor
A tensor representation of the pixel data.
- ImageFrame:to_float_tensor ()
-
Copies video frame pixel data into a
torch.FloatTensor.For most pixel formats, this means that all channels will contain values between 0 and 1. If the pixel format is YUV, the chroma channels will contain values between -1 and 1.
Returns:
-
torch.FloatTensor
A tensor representation of the pixel data.
- ImageFrame:timestamp ()
-
Get the timestamp of this image frame (in seconds).
Returns:
-
number
The timestamp (in seconds).
Class Video
- Video.new (path)
-
Creates a new Video.
Parameters:
- path string Absolute or relative path to a video file.
Returns:
- Video:duration ()
-
Get the duration of the video in seconds.
Returns:
-
number
The duration of the video in seconds.
- Video:guess_image_frame_rate ()
-
Guess the average frame rate of the video in FPS.
Returns:
-
number
The frame rate, or zero if unknown.
- Video:get_image_frame_count ()
-
Get the number of image frames in the video.
Returns:
-
number
The number of frames, or zero if unknown.
- Video:filter (pixel_format_name[, filterchain='null'])
-
Apply a filterchain to the video.
Parameters:
- pixel_format_name string The desired output pixel format.
- filterchain string A description of the filterchain. (default 'null')
Returns:
-
number
The duration of the video in seconds.
- Video:next_image_frame ()
-
Read the next video frame from the video.
Returns:
- Video:seek (seek_target)
-
Seek to the first keyframe before the frame number specified.
Parameters:
- seek_target number The position to seek to (in seconds).
Returns:
-
Video
This video object.