No. It is designed as an operator-support system. Alerts and risk cues assist decision-making, but final judgment remains with human operators. (Tam otonom değil; operatör destek)

False alarms are reduced by combining detection confidence with temporal behavior (tracking) and explainable Threat Score logic rather than single-frame triggers. (Tek kare değil, zaman + skor)

Night performance is challenged by low contrast and noise. The project evaluates low-light/thermal-like scenarios and uses temporal consistency to stabilize risk cues. (Gece zor; tutarlılık ile stabilize)

No. It prioritizes human-in-the-loop workflows: the UI is built to clarify, not to automate enforcement. (Operatör yerine geçmez)

No. The dataset strategy explicitly avoids sensitive real-world border footage. Only public/open-source/synthetic sources are used. (Hassas gerçek sınır verisi yok)