Refinement Co‐supervision network for real‐time semantic segmentation

Abstract

Semantic segmentation is a fundamental technology for autonomous driving. It has a high demand for inference speed and accuracy. However, a good trade‐off between accuracy and latency is yet not present in existing semantic segmentation approaches. Due to the limitation of speed, the authors cannot increase the number of network layers without limit and cannot design modules like in the networks without real‐time. It is a challenging problem how to design a model with good performance under limited resources. To alleviate these issues, in this study, the authors propose a refinement co‐supervision network (RCNet), which is real‐time on a high‐resolution image (1024×2048). The authors first construct the context refinement module, which can provide low computation cost way for obtaining the large receptive field and context information. Furthermore, a boundary co‐supervision mechanism is proposed. It strengthens the optimisation of easily neglected boundaries and small targets. Experimental results reveal that the proposed RCNet outperforms seven representative semantic segmentation methods.

Multi-Scale Feature Selective Matching Network for Object Detection

“…By Mask R-CNN [28] found pixel bias in the RoI Pooling layer, so it used bilinear interpolation to replace the RoI Pooling layer with the RoI Align layer to achieve better detection. In addition, the mask head uses a top-down approach for segmentation [29][30][31]. The use of region proposals with appropriate ORPs for training increased the detection performance of irregular objects in [32].…”

Multiple space based cascaded center point network for object detection

“…To alleviate the imbalance problem in detecting different scale objects, many methods have been put out, such as enhancing object feature information, improving training strategies, data expansion and so on [19]. These methods have better detection results [20][21][22]. Aiming at the phenomenon of sparse fault samples in actual industrial environments, a method of fault diagnosis with few samples based on parameter optimization and feature metrics is proposed [23].…”