论文标题
部分可观测时空混沌系统的无模型预测
SEA: Bridging the Gap Between One- and Two-stage Detector Distillation via SEmantic-aware Alignment
论文作者
论文摘要
我们重新审视单阶段检测器蒸馏任务，并提出一个简单有效的语义感知框架，以填补它们之间的空白。我们通过设计类别的锚点来解决每个类别的代表性模式，并将像素和类别锚之间的拓扑距离正规化以进一步拧紧其语义键，以解决每个类别的代表性模式，以解决像素级级别的不平衡问题。鉴于语义依赖于良好的蒸馏功效，我们将方法的海洋（语义意识对齐）蒸馏列为蒸馏。 SEA非常适合检测管道，并在一个挑战性的可可对象检测任务上获得了新的最先进的结果。它在实例分割上的出色性能进一步体现了概括能力。带有Resnet50-FPN的2倍延伸视网膜和FCO的表现分别优于其相应的3X Resnet101-FPN老师，分别到达40.64和43.06 AP。代码将公开可用。
We revisit the one- and two-stage detector distillation tasks and present a simple and efficient semantic-aware framework to fill the gap between them. We address the pixel-level imbalance problem by designing the category anchor to produce a representative pattern for each category and regularize the topological distance between pixels and category anchors to further tighten their semantic bonds. We name our method SEA (SEmantic-aware Alignment) distillation given the nature of abstracting dense fine-grained information by semantic reliance to well facilitate distillation efficacy. SEA is well adapted to either detection pipeline and achieves new state-of-the-art results on the challenging COCO object detection task on both one- and two-stage detectors. Its superior performance on instance segmentation further manifests the generalization ability. Both 2x-distilled RetinaNet and FCOS with ResNet50-FPN outperform their corresponding 3x ResNet101-FPN teacher, arriving 40.64 and 43.06 AP, respectively. Code will be made publicly available.
