MetaGraspNet
收藏资源简介:
MetaGraspNet数据集是由滑铁卢大学视觉与图像处理研究组创建,用于多任务学习的复杂视觉任务测试。该数据集因其庞大的数据量、增加的任务复杂性和高现实价值而被用作新的测试平台。数据集的创建旨在解决多任务学习中的挑战,特别是在使用轻量级特征提取骨干网络时的效率问题。MetaGraspNet数据集的应用领域包括机器人视觉和计算机视觉,旨在通过多任务学习提高模型在多个任务上的性能。
MetaGraspNet Dataset was developed by the Vision and Image Processing Research Group of the University of Waterloo, serving as a testbed for complex visual task evaluation in multi-task learning. Featuring its massive data scale, increased task complexity and high practical value, this dataset has been employed as a novel test platform. The dataset is created to address the challenges in multi-task learning, particularly the efficiency issues when using lightweight feature extraction backbone networks. The application domains of MetaGraspNet Dataset cover robot vision and computer vision, aiming to improve the model's performance across multiple tasks via multi-task learning.




