<p>Indoor object detection in real-world deployments often suffers from severe performance degradation due to complex structural corruptions such as motion blur, sensor noise, and compression artifacts, which are exacerbated by low-resolution capture in resource-constrained settings. This performance drop poses a critical challenge for applications in the interior design sector, including virtual home staging, intelligent furniture recognition, and AR/VR interior design experiences. Existing solutions either rely on computationally expensive super-resolution (SR) preprocessing pipelines or employ degradation-agnostic knowledge distillation (KD), the latter being limited by its inability to address different degradation types in a targeted manner. In this work, we introduce progressive degradation-aware distillation (PDAD), a zero-inference-overhead training framework that equips detectors with strong robustness against diverse degradations. PDAD first learns a degradation-aware proxy that captures the degradation characteristics of the input, and then uses this proxy as conditional guidance to adaptively refine detection features. All auxiliary components are discarded after training, leaving the deployed model identical in architecture and speed to its baseline counterpart. Extensive experiments on indoor datasets such as HomeObjects-3K show that PDAD achieves substantial gains over state-of-the-art (SOTA) methods under challenging mixed degradations and generalizes well to real-world indoor imagery. This robust capability is particularly crucial for real-world applications in interior design, such as automated inventory, space layout planning, and AR-based visualization. By providing a low-cost yet effective solution for robust indoor object detection, PDAD facilitates the widespread adoption of AI in resource-constrained design settings.</p>