In the cigarette manufacturing industry, filter bag damage within baghouse dust collectors can lead to dust leakage, posing serious threats to production safety and the surrounding environment. Traditional manual inspection methods are inefficient and struggle to achieve real-time, precise detection. To address this, a filter bag damage detection method, IFB-YOLO, based on an improved YOLOv11, is proposed. Addressing the multi-scale nature, irregular morphology, and blurred boundaries of filter bag damage, this method incorporates the PKIBlock module within C3k2. This leverages parallel multi-scale deep convolutions to enhance texture feature extraction for minute and diverse damage patterns. An LSKA module is integrated post-backbone network to strengthen the model is ability to focus on critical damage areas within complex backgrounds. The MPDIoU loss function is employed to optimise damage localisation accuracy by minimising diagonal distances between bounding box corners. Experiments on dust collector filter bag damage datasets demonstrate that IFB-YOLO achieves a mean average precision at 50% (mAP50) of 95.3%, representing a 2.8% improvement over baseline models and outperforming other mainstream detection approaches. These results validate the feasibility and effectiveness of the enhanced model.