Automated Video Scoring Using Physiological Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current visual feature extraction scoring systems for interview videos lack efficiency in analyzing and scoring non-verbal physiological features, such as head postures and eye gazes, which are crucial for assessing candidate performance during interviews.
Innovation Solution
The system extracts physiological features from video clips using data processors to generate visual words, which are then converted into feature vectors and used in a regression model for scoring, incorporating algorithms like doc2vec and machine learning models to predict personality traits and hiring decisions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If visual feature extraction is used to analyze physiological features in interview videos, then automated scoring capability is improved, but analysis efficiency and processing speed deteriorate
Solution Approach 1:
The system segments the interview video into individual frames and extracts physiological features frame-by-frame. Each frame is processed independently to identify head postures, eye gazes, and facial expressions, converting continuous video data into discrete analyzable units that improve processing efficiency
Solution Approach 2:
The patent introduces visual words as an intermediary representation between raw physiological features and final scoring. Extracted features are converted into visual word documents, which then feed into vector generation algorithms and regression models, creating a multi-stage processing pipeline that balances automation with efficiency
2Measurement precision
If comprehensive physiological features are extracted from each video frame, then measurement precision of candidate behavior is improved, but computational complexity and processing time worsen
Solution Approach 1:
The system extracts only the most relevant physiological features from each video frame, specifically focusing on head postures, eye gazes, and facial expressions. This selective extraction approach captures essential behavioral information while avoiding unnecessary computational overhead from analyzing all visual elements
Solution Approach 2:
The patent transforms continuous physiological measurements into discrete visual words through clustering algorithms. By converting continuous feature data into categorical visual word representations, the system reduces computational complexity while preserving the essential information needed for accurate scoring
Data Source
AI summary
Systems and methods are provided for scoring video clips using visual feature extraction. A signal including a video clip of a subject is received. For each frame of the video clip, physiological features of the subject visually rendered in the video clip are extracted. A plurality of visual words associated with the extracted physiological features are determined. A document including the plurality of visual words is generated. A plurality of feature vectors associated with the document are determined. The plurality of feature vectors to a regression model for scoring are provided.