The paper introduces Richer Convolutional Features (RCF), a novel deep learning architecture designed to enhance edge detection in images. RCF leverages the rich feature hierarchies from all convolutional layers of a pre-trained VGG16 network, capturing both semantic and fine detail information. This approach aims to overcome the limitations of existing methods that often fail to capture complex data structures due to variations in scale and aspect ratios. RCF is trained using backpropagation and achieves state-of-the-art performance on several datasets, including the BSDS500 benchmark, with an ODS F-measure of 0.811 and a fast speed of 8 FPS. A fast version of RCF, achieving an ODS F-measure of 0.806 at 30 FPS, is also presented. The versatility of RCF is demonstrated by its application in classical image segmentation tasks, where it outperforms recent edge detectors and traditional methods. The paper discusses the architecture's design, loss function, multiscale hierarchical edge detection, and comparisons with other methods, highlighting its effectiveness and efficiency.The paper introduces Richer Convolutional Features (RCF), a novel deep learning architecture designed to enhance edge detection in images. RCF leverages the rich feature hierarchies from all convolutional layers of a pre-trained VGG16 network, capturing both semantic and fine detail information. This approach aims to overcome the limitations of existing methods that often fail to capture complex data structures due to variations in scale and aspect ratios. RCF is trained using backpropagation and achieves state-of-the-art performance on several datasets, including the BSDS500 benchmark, with an ODS F-measure of 0.811 and a fast speed of 8 FPS. A fast version of RCF, achieving an ODS F-measure of 0.806 at 30 FPS, is also presented. The versatility of RCF is demonstrated by its application in classical image segmentation tasks, where it outperforms recent edge detectors and traditional methods. The paper discusses the architecture's design, loss function, multiscale hierarchical edge detection, and comparisons with other methods, highlighting its effectiveness and efficiency.