Convolutional neural networks have made significant progress in edge detection by progressively exploring the context and semantic features. However, most methods struggle to produce pixel-accurate edge maps and suffer from false positives in the high-frequency texture regions of non-edge part. In this paper, We propose a new edge detection method that aims to adjust the model’s focus on different pixels in the non-edge area based on textureness and enhance the accuracy of edge detection. Specifically, we first propose a weighting strategy based on textureness and obtain a textureness-aware loss RWCE, which can guide the model to pay more attention to the learning of high-frequency texture regions during the training process, thus improving the prediction accuracy of these regions. Moreover, we design an end-to-end network which adopts the bottom-up/top-down architecture, effectively utilizing hierarchical features, progressively increasing the resolution of feature maps, and ultimately generating pixel-accurate edge maps. Our method achieves promising performance on BSDS500, BIPED, NYUD, and outperforming most previous methods. The source code of this work is available at: https://github.com/yx-yyds/TANet .