Video displaying method and apparatus, device and computer storage medium

Patent metadata

Field	Value
Publication number	US-10817149-B2
Application number	US-201816183464-A
Country	US
Kind code	B2
Filing date	Nov 7, 2018
Priority date	Nov 28, 2017
Publication date	Oct 27, 2020
Grant date	Oct 27, 2020

Abstract

The present disclosure provides a video displaying method and apparatus, a device and a computer storage medium, wherein the method comprises: positioning a target video; displaying a video preview of the target video, the video preview being a video segment extracted from the target video. This displaying manner exhibits a stronger capability of expressing the video content. The user can better know whether the video is the desired content. The present disclosure may reduce a probability of the user missing desired video resources and reduce a probability that the user finds that the video content does not meet an expectation after he views the video content, thereby helping the user to conveniently and effectively view the desired video resources.

First claim

What is claimed is:

1. A video displaying method, wherein the method comprises:
positioning a target video;
displaying a video preview of the target video, the video preview being a video segment extracted from the target video by:
inputting the target video into a video classification model obtained by pre-training;
obtaining weights of respective video frames output by an attention module in the video classification model; and
extracting continuous N video frames whose total weight value satisfies a preset requirement, as the video preview of the target video, N being a preset positive integer,

wherein a process of training the video classification model comprises:
regarding a video whose video class is pre-annotated as training data;
training the video classification model by regarding the video of the training data as input of the video classification model and by regarding the corresponding video class as output of the video classification model, to minimize a loss function of a classification result, and
wherein during the training of the video classification model, regarding the video in the training data as input of a convolutional neural network, and outputting convolutional features of respective frames in the video;
regarding the convolutional features of respective frames as input of another neural network to output time sequence features of respective frames;
regarding time sequence features of respective frames as input of an attention module to output weights of respective frames;
mapping to a video type at a fully-connected layer according to weights of respective frames and output of the another neural network;
using a mapping result to calculate a loss function
wherein the another neural network comprise one of a Long Short-Term Memory (LSTM), a Gated Recurrent Unit (GRU) or a Recurrent Neural Network (RNN).

2. The method according to claim 1 , wherein the positioning a target video comprises:
positioning a video at a target position in a video feed page; or
positioning a video at a target position in a video aggregation page.

3. The method according to claim 1 , wherein the displaying a video preview of the target video comprises:
after positioning the target video, automatically playing the video preview of the target video; or
playing the video preview of the target video after detecting that a user triggers an event for playing the video preview.

4. The method according to claim 3 , wherein during the playing of the video preview of the target video, display prompt information of the video preview which is being played.

5. The method according to claim 3 , wherein the playing the video preview of the target video comprises:
fast playing the video preview of the video.

6. The method according to claim 1 , wherein the method further comprises:
playing the target video after detecting that a user triggers an event for playing the target video.

7. The method according to claim 6 , wherein the playing the target video after detecting that the user triggers the event for playing the target video comprises:
accessing a playing link of the target video to which a specific component is directed, after detecting the user triggers a specific component on a page of displaying the video preview.

8. A device, wherein the device comprises:
one or more processors,
a storage for storing one or more programs,
the one or more programs, when executed by said one or more processors, enable said one or more processors to implement a video displaying method, wherein the method comprises:
positioning a target video;
displaying a video preview of the target video, the video preview being a video segment extracted from the target video by:
inputting the target video into a video classification model obtained by pre-training;
obtaining weights of respective video frames output by an attention module in the video classification model; and
extracting continuous N video frames whose total weight value satisfies a preset requirement, as the video preview of the target video, N being a preset positive integer,

wherein a process of training the video classification model comprises:
regarding a video whose video class is pre-annotated as training data;
training the video classification model by regarding the video of the training data as input of the video classification model and by regarding the corresponding video class as output of the video classification model, to minimize a loss function of a classification result, and
wherein during the training of the video classification model, regarding the video in the training data as input of a convolutional neural network, and outputting convolutional features of respective frames in the video;
regarding the convolutional features of respective frames as input of another neural network to output time sequence features of respective frames;
regarding time sequence features of respective frames as input of an attention module to output weights of respective frames;
mapping to a video type at a fully-connected layer according to weights of respective frames and output of the another neural network;
using a mapping result to calculate a loss function
wherein the another neural network comprise one of a Long Short-Term Memory (LSTM), a Gated Recurrent Unit (GRU) or a Recurrent Neural Network (RNN).

9. The device according to claim 8 , wherein the positioning a target video comprises:
positioning a video at a target position in a video feed page; or
positioning a video at a target position in a video aggregation page.

10. The device according to claim 8 , wherein the displaying a video preview of the target video comprises:
after positioning the target video, automatically playing the video preview of the target video; or
playing the video preview of the target video after detecting that a user triggers an event for playing the video preview.

11. The device according to claim 10 , wherein during the playing of the video preview of the target video, display prompt information of the video preview which is being played.

12. The device according to claim 10 , wherein the playing the video preview of the target video comprises:
fast playing the video preview of the video.

13. The device according to claim 8 , wherein the method further comprises:
playing the target video after detecting that a user triggers an event for playing the target video.

14. The device according to claim 13 , wherein the playing the target video after detecting that the user triggers the event for playing the target video comprises:
accessing a playing link of the target video to which a specific component is directed, after detecting the user triggers a specific component on a page of displaying the video preview.

15. A non-transitory storage medium including computer-executable instructions which, when executed by a computer processor, execute a video displaying method, wherein the method comprises:
positioning a target video;
displaying a video preview of the target video, the video preview being a video segment extracted from the target video by:
inputting the target video into a video classification model obtained by pre-training;
obtaining weights of respective video frames output by an attention module in the video classification model; and
extracting continuous N video frames whose total weight value satisfies a preset requirement, as the video preview of the target video, N being a preset positive integer,

wherein a process of training the video classification model comprises:
regarding a video whose video class is pre-ann

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Yu Xinwei
• Zeng Xi
• Li Fu
• Wen Shilei
• Jiang Jiaqi
• Zhang Xin
• Li Feng
• Zhang Baoyu
• Qi Meng
• Gong Lixing
• Han Rongrong
• Song Yueyang
• Meng Fantian
• Wang Haiping
• Liu Xiao
• Sun Hao
• Li Xubin
• Chang Chenghao

Classifications

• G06V20/46

  Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames · CPC title

• H04N21/8549 Primary

  Creating video summaries, e.g. movie trailer {(retrieval in video databases by using presentations in form of a video summary G06F16/739)} · CPC title

• G06F3/0484 Primary

  for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range · CPC title

• G06F18/241

  relating to the classification model, e.g. parametric or non-parametric approaches · CPC title

• G06F18/2148

  characterised by the process organisation or structure, e.g. boosting cascade · CPC title