Video Format Classification Using Convolutional Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video sharing platforms require manual user intervention to upload and transcode spherical videos, limiting the upload process for mobile users and lacking support for multiple video formats, making it non-intuitive and inefficient.
Innovation Solution
A fully-automatic system using a supervised convolutional neural network to classify video formats, such as 2D, equirectangular 360, and equirectangular 360 stereo, by pre-processing video frames, predicting confidence values, determining the spatial format, and adding necessary metadata for correct transcoding and display.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual user selection is required to identify video format, then users can specify the correct format, but the upload process becomes complex and non-intuitive
Solution Approach 1:
The system performs automatic video format classification without requiring user intervention. The convolutional neural network analyzes video frames and autonomously determines whether the video is 2D, equirectangular 360, or equirectangular 360 stereo, eliminating the need for manual format selection while maintaining accurate format identification.
Solution Approach 2:
The video format classification is performed automatically during the upload process before transcoding and rendering. The system pre-processes video frames, runs them through the neural network classifier, and obtains format predictions in advance, allowing the platform to prepare appropriate transcoding parameters and metadata before actual processing begins.
2Adaptability or versatility
If multiple video formats are supported, then the platform can handle diverse content, but the system complexity increases
Solution Approach 1:
The video format classification problem is segmented into distinct categories (2D, equirectangular 360, equirectangular 360 stereo) with dedicated processing paths. The convolutional neural network is trained to recognize specific visual characteristics of each format type, allowing the system to handle multiple formats through specialized classification rather than a single complex processing pipeline.
Solution Approach 2:
The convolutional neural network acts as an intermediary between video upload and transcoding/rendering processes. It automatically classifies video formats and generates predictions that guide subsequent processing steps, eliminating the need for complex manual configuration or multiple specialized processing systems for each format type.
3Extent of automation
If automatic video format classification is implemented, then user intervention is eliminated, but processing time increases
Solution Approach 1:
The system extracts and processes only a subset of video frames rather than analyzing the entire video. By selecting representative frames for classification, the neural network can determine video format quickly without the computational overhead of processing all frames, significantly reducing processing time while maintaining classification accuracy.
Solution Approach 2:
The essential format-classifying features are extracted from video frames before neural network processing. The system identifies and extracts key visual characteristics that distinguish different video formats, removing unnecessary data and focusing computational resources only on the features needed for accurate classification, thereby reducing overall processing time.
Data Source
AI summary
Methods, systems, and computer program products for classifying a spatial format of a video file. The system includes one or more processors and a memory coupled to the processors. The memory stores data comprising program code that, when executed by the processors, causes the system to allow video sharing platforms to support multiple formats without asking users to manually identify the format of the video.