Research on the dynamic capture and analysis system of symphony stage performance based on deep convolutional neural network (DCNN) and computer vision technology

The analysis of symphony stage performances requires capturing intricate details such as conductor gestures, musician movements, and interactions between instrumental sections. Manual observation or basic video analysis is not effective because it is imprecise for dynamic performance, while the present method is accurate and efficient. This work assesses the establishment of a capturing and analysing system for performances of symphonies utilizing Deep Convolutional Neural Networks (DCNN) and computer vision. As this research utilises secondary data, it employs a quantitative strategy, and its data sources are high-quality recordings of famous orchestras. The DCNN model identifies several crucial visual features facilitating the identification of performance elements such as tempo change, musicians’ synchronization, and conductor’s cues. It is established that the DCNN-based system impressively improves classification accuracy scoring $\mathbf{8 7. 3 \%}$ in performance dynamics detection compared to existing methods. In support of the model, percentages like precision $(\mathbf{8 4. 1 \%})$ as well as recall ($\mathbf{7 9. 6 \%)}$) support the model’s efficiency. The paper demonstrates a new way of using more intelligent AI with computer vision in achieving new and better ways of analysing performances, especially as relates to symphonies.