Article: Real-time teaching behaviour recognition and ability dynamic evaluation technology based on image sequence mining Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.49 pp.140 - 159 Abstract: How to achieve objective teaching process identification and real-time teacher ability evaluation has become an important direction of educational informatisation research. This paper puts forward a model of teaching behaviour recognition and ability dynamic evaluation, which combines 3D-CNN and Bi-LSTM, and realises automatic recognition of classroom teaching behaviour and continuous quantitative analysis of teachers' ability through visual data. The experimental results show that the proposed model achieves 93.6% accuracy, 91.7% recall and 0.92 F1-value in the task of teaching behaviour recognition, which is significantly better than the traditional convolution or single time series model. The dynamic evaluation module realises the real-time tracking of teachers' ability through the time-weighted mechanism, and the coincidence rate between the system and the manual expert score reaches 91.8%. The research results finds this method boosts the intelligent level of teaching process monitoring, promote the transformation from subjective experience evaluation to data-driven dynamic evaluation. Inderscience Publishers - linking academia, business and industry through research

Title: Real-time teaching behaviour recognition and ability dynamic evaluation technology based on image sequence mining

Authors: Juan Li

Addresses: School of Science College (Normal College), Hunan Shaoyang University, Shaoyang, 422000, Hunan, China

Abstract: How to achieve objective teaching process identification and real-time teacher ability evaluation has become an important direction of educational informatisation research. This paper puts forward a model of teaching behaviour recognition and ability dynamic evaluation, which combines 3D-CNN and Bi-LSTM, and realises automatic recognition of classroom teaching behaviour and continuous quantitative analysis of teachers' ability through visual data. The experimental results show that the proposed model achieves 93.6% accuracy, 91.7% recall and 0.92 F1-value in the task of teaching behaviour recognition, which is significantly better than the traditional convolution or single time series model. The dynamic evaluation module realises the real-time tracking of teachers' ability through the time-weighted mechanism, and the coincidence rate between the system and the manual expert score reaches 91.8%. The research results finds this method boosts the intelligent level of teaching process monitoring, promote the transformation from subjective experience evaluation to data-driven dynamic evaluation.

Keywords: image sequence mining; teaching behaviour recognition; dynamic capability evaluation.

DOI: 10.1504/IJICT.2025.151065

International Journal of Information and Communication Technology, 2025 Vol.26 No.49, pp.140 - 159

Received: 14 Oct 2025
Accepted: 11 Nov 2025
Published online: 12 Jan 2026 *

Title: Real-time teaching behaviour recognition and ability dynamic evaluation technology based on image sequence mining

Keep up-to-date