Driven by Convolutional Neural Networks, object detection and semantic segmentation have gained significant improvements. However, existing methods on the basis of a full top-down module have limited robustness in handling those two tasks simultaneously. To this end, we present a joint multi-task framework, termed IvaNet. Different from existing methods, our IvaNet backwards abstract semantic information from higher layers to augment lower layers using local top-down modules. The comparisons against some counterparts on the PASCAL VOC and MS COCO datasets demonstrate the functionality of IvaNet.
翻译:在进化神经网络、物体探测和语义分割的驱动下,在进化神经网络、物体探测和语义分割的驱动下,已经取得了显著的改进,然而,基于一个完整的自上而下模块的现有方法在同时处理这两项任务方面力度有限。为此,我们提出了一个称为 IvaNet 的联合多任务框架。不同于现有方法,我们的IvaNet从高层向后退抽象语义信息,用本地自上而下模块扩大下层。与PASCAL VOC 和 MS COCO 数据集的一些对口单位的比较显示了IvaNet 的功能。