Yang Defu, Solihin Mahmud Iwan, Zhao Yawen, Cai Bingyu, Chen Chaoran, Wijaya Andika Aji, Ang Chun Kit, Lim Wei Hong
Faculty of Engineering, Technology and Built Environment, UCSI University, Kuala Lumpur, Malaysia.
School of Advanced Manufacturing, Shantou Polytechnic, Shantou, China.
iScience. 2024 Dec 17;28(1):111618. doi: 10.1016/j.isci.2024.111618. eCollection 2025 Jan 17.
Achieving lightweight real-time object detection necessitates balancing model compression with detection accuracy, a difficulty exacerbated by low redundancy and uneven contributions from convolutional layers. As an alternative to traditional methods, we propose Rigorous Gradation Pruning (RGP), which uses a desensitized first-order Taylor approximation to assess filter importance, enabling precise pruning of redundant kernels. This approach includes the iterative reassessment of layer significance to protect essential layers, ensuring effective detection performance. We applied RGP to YOLOv8 detectors and tested it on GTSDB, Seaships, and COCO datasets. On GTSDB, RGP achieved 80% compression of YOLOv8n with only a 0.11% drop in mAP0.5, while increasing frames per second (FPS) by 43.84%. For YOLOv8x, RGP achieved 90% compression, a 1.26% mAP0.5:0.95 increase, and a 112.66% FPS boost. Significant compression was also achieved on Seaships and COCO datasets, demonstrating RGP's robustness across diverse object detection tasks and its potential for advancing efficient, high-speed detection models.
要实现轻量级实时目标检测，需要在模型压缩和检测精度之间取得平衡，而卷积层的低冗余度和贡献不均加剧了这一难题。作为传统方法的替代方案，我们提出了严格梯度剪枝（RGP），它使用脱敏的一阶泰勒近似来评估滤波器的重要性，从而能够精确地剪枝冗余内核。这种方法包括对层重要性进行迭代重新评估，以保护关键层，确保有效的检测性能。我们将RGP应用于YOLOv8检测器，并在GTSDB、Seaships和COCO数据集上进行了测试。在GTSDB上，RGP实现了对YOLOv8n 80%的压缩，mAP0.5仅下降0.11%，同时每秒帧数（FPS）提高了43.84%。对于YOLOv8x，RGP实现了90%的压缩，mAP0.5:0.95提高了1.26%，FPS提高了112.66%。在Seaships和COCO数据集上也实现了显著的压缩，证明了RGP在各种目标检测任务中的鲁棒性及其推进高效、高速检测模型的潜力。
