HF-YOLO is an advanced pedestrian detection model that addresses challenges such as scale variations and occlusions. It improves detection by integrating shallow localization information and deep semantic information through feature fusion, adding a high-resolution detection layer, and using the HardSwish activation function to enhance feature representation. A balance factor is introduced to the CIoU loss function to resolve regression imbalance, improving pedestrian localization accuracy. Experimental results show that HF-YOLO achieves a 3.52% increase in average precision, a 1.35% boost in accuracy, and a 4.83% enhancement in recall, with real-time performance of 8.5ms. The model outperforms existing approaches in detection accuracy and reliability. HF-YOLO incorporates a novel feature fusion module, the HardSwish activation function, and the F-CIoU loss function to enhance performance in handling diverse object scales and occluded pedestrians. The model maintains real-time performance and achieves improved detection accuracy. The paper presents the model structure, experimental setup, and results, demonstrating the effectiveness of HF-YOLO in pedestrian detection.HF-YOLO is an advanced pedestrian detection model that addresses challenges such as scale variations and occlusions. It improves detection by integrating shallow localization information and deep semantic information through feature fusion, adding a high-resolution detection layer, and using the HardSwish activation function to enhance feature representation. A balance factor is introduced to the CIoU loss function to resolve regression imbalance, improving pedestrian localization accuracy. Experimental results show that HF-YOLO achieves a 3.52% increase in average precision, a 1.35% boost in accuracy, and a 4.83% enhancement in recall, with real-time performance of 8.5ms. The model outperforms existing approaches in detection accuracy and reliability. HF-YOLO incorporates a novel feature fusion module, the HardSwish activation function, and the F-CIoU loss function to enhance performance in handling diverse object scales and occluded pedestrians. The model maintains real-time performance and achieves improved detection accuracy. The paper presents the model structure, experimental setup, and results, demonstrating the effectiveness of HF-YOLO in pedestrian detection.