CVPR2020:基于自适应采样的非局部神经网络鲁棒点云处理(PointASNL) PointASNL: Robust Point Clouds Processing Using Nonlocal Neural Networks With Adaptive Sampling 论文地址: https://openaccess.thecvf.com/content_CVPR_2020/html/Yan_PointASNL_Robust_Point_Clouds_Processing_Using_No
论文题目:Beyond OCR + VQA: Involving OCR into the Flow for Robust and Accurate TextVQA 论文链接:https://dl.acm.org/doi/abs/10.1145/3474085.3475606 一.任务概述 视觉问答任务(VQA):将图像和关于图像的自然语言问题作为输入,并生成自然语言答案作为输出. 文本视觉问答任务(TextVQA):面向文字识别的问答任务. 二.Baseline 2.1 Baseline 1