射擊評分更為公正：SingTown科技視覺AI自動評分系統

SingTown基於OpenMV嘅智能相機解決方案，能夠實時識別同精準定位圓形目標（目標環同靶心），用於射擊訓練同競賽計分。系統利用AI圖像識別同機器視覺演算法，以高幀率捕捉目標表面變化，自動標註靶心坐標，並生成命中分佈同分數統計，為教練評估、訓練改進同模擬射擊場提供可靠數據支援。

In training and competitive scenarios, the core requirements for enhancing training efficiency and ensuring fairness are the rapid and accurate identification of the bullseye on circular targets and the generation of hit scores. SingTown Technology's intelligent camera solution, based on the OpenMV platform, applies edge AI and computer vision technology to target recognition and scoring analysis, achieving stable and real-time hit detection under various lighting and target material conditions.

The system utilises high-speed image acquisition to detect the target surface in real time, employing image segmentation and edge detection algorithms to accurately extract annular contours. It computes the geometric centres and coordinates of the target rings and bullseye, and performs pixel-level localisation of the impact point (or simulated hit point) of incoming projectiles within the frame.

Based on the recognition results, the machine vision module generates hit heatmaps, hit distribution statistics, and scoring curves for single/multiple shots. It supports data aggregation by target position, shooter, or time period, facilitating instructors in quantitatively assessing shooter stability, deviation trends, and training progress. The system also supports multi-camera synchronous calibration, achieving full-range coverage and spatiotemporal comparison of the shooting range, thereby ensuring scoring consistency and traceability.

AI models can undergo customised training based on the reflective characteristics of the target, printing variations on the target surface, and ambient lighting to enhance robustness across different scenarios. Recognition and scoring data can be exported as CSV files or visualised reports, facilitating training archive management and subsequent analysis.

Moreover, the OpenMV intelligent camera offers advantages such as low power consumption, compact size, and high frame rate for edge deployment, enabling easier and more flexible integration into mobile target platforms, fixed target walls, or simulation training devices.

SingTown將圖像識別與機器視覺深度整合至訓練系統中，將射擊場的視覺評分系統升級為可量化、可複查的智能評估平台。這顯著提升了訓練效率與評分公正性，為教育、體育及模擬訓練場景提供專業的嵌入式視覺評分解決方案。