• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권3호
• /
• pp.861-880
• /
• 2023

Semantic segmentation of road scene is the key technology of autonomous driving, and the improvement of convolutional neural network architecture promotes the improvement of model segmentation performance. The existing convolutional neural network has the simplification of learning knowledge and the complexity of the model. To address this issue, we proposed a road scene semantic segmentation algorithm based on multi-task collaborative learning. Firstly, a depthwise separable convolution atrous spatial pyramid pooling is proposed to reduce model complexity. Secondly, a collaborative learning framework is proposed involved with saliency detection, and the joint loss function is defined using homoscedastic uncertainty to meet the new learning model. Experiments are conducted on the road and nature scenes datasets. The proposed method achieves 70.94% and 64.90% mIoU on Cityscapes and PASCAL VOC 2012 datasets, respectively. Qualitatively, Compared to methods with excellent performance, the method proposed in this paper has significant advantages in the segmentation of fine targets and boundaries.

• ETRI Journal
• /
• 제45권5호
• /
• pp.822-835
• /
• 2023

Recent semantic segmentation frameworks usually combine low-level and high-level context information to achieve improved performance. In addition, postlevel context information is also considered. In this study, we present a Context ReFinement Network (CRFNet) and its training method to improve the semantic predictions of segmentation models of the encoder-decoder structure. Our study is based on postprocessing, which directly considers the relationship between spatially neighboring pixels of a label map, such as Markov and conditional random fields. CRFNet comprises two modules: a refiner and a combiner that, respectively, refine the context information from the output features of the conventional semantic segmentation network model and combine the refined features with the intermediate features from the decoding process of the segmentation model to produce the final output. To train CRFNet to refine the semantic predictions more accurately, we proposed a sequential training scheme. Using various backbone networks (ENet, ERFNet, and HyperSeg), we extensively evaluated our model on three large-scale, real-world datasets to demonstrate the effectiveness of our approach.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제16권1호
• /
• pp.60-79
• /
• 2022

The U-Net architecture-based segmentation models attained remarkable performance in numerous medical image segmentation missions like skin lesion segmentation. Nevertheless, the resolution gradually decreases and the loss of spatial information increases with deeper network. The fusion of adjacent layers is not enough to make up for the lost spatial information, thus resulting in errors of segmentation boundary so as to decline the accuracy of segmentation. To tackle the issue, we propose a new deep learning-based segmentation model. In the decoding stage, the feature channels of each decoding unit are concatenated with all the feature channels of the upper coding unit. Which is done in order to ensure the segmentation effect by integrating spatial and semantic information, and promotes the robustness and generalization of our model by combining the atrous spatial pyramid pooling (ASPP) module and channel attention module (CAM). Extensive experiments on ISIC2016 and ISIC2017 common datasets proved that our model implements well and outperforms compared segmentation models for skin lesion segmentation.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권8호
• /
• pp.1996-2015
• /
• 2023

High-definition (HD) maps can provide precise road information that enables an autonomous driving system to effectively navigate a vehicle. Recent research has focused on leveraging semantic segmentation to achieve automatic annotation of HD maps. However, the existing methods suffer from low recognition accuracy in automatic driving scenarios, leading to inefficient annotation processes. In this paper, we propose a novel semantic segmentation method for automatic HD map annotation. Our approach introduces a new encoder, known as the convolutional transformer hybrid encoder, to enhance the model's feature extraction capabilities. Additionally, we propose a multi-level fusion module that enables the model to aggregate different levels of detail and semantic information. Furthermore, we present a novel decoupled boundary joint decoder to improve the model's ability to handle the boundary between categories. To evaluate our method, we conducted experiments using the Bird's Eye View point cloud images dataset and Cityscapes dataset. Comparative analysis against stateof-the-art methods demonstrates that our model achieves the highest performance. Specifically, our model achieves an mIoU of 56.26%, surpassing the results of SegFormer with an mIoU of 1.47%. This innovative promises to significantly enhance the efficiency of HD map automatic annotation.

• Journal of Information Processing Systems
• /
• 제17권5호
• /
• pp.892-904
• /
• 2021

An image semantic segmentation model is proposed based on improved ENet network in order to achieve the low accuracy of image semantic segmentation in complex environment. Firstly, this paper performs pruning and convolution optimization operations on the ENet network. That is, the network structure is reasonably adjusted for better results in image segmentation by reducing the convolution operation in the decoder and proposing the bottleneck convolution structure. Squeeze-and-excitation (SE) module is then integrated into the optimized ENet network. Small-scale targets see improvement in segmentation accuracy via automatic learning of the importance of each feature channel. Finally, the experiment was verified on the public dataset. This method outperforms the existing comparison methods in mean pixel accuracy (MPA) and mean intersection over union (MIOU) values. And in a short running time, the accuracy of the segmentation and the efficiency of the operation are guaranteed.

• 한국인터넷방송통신학회논문지
• /
• 제19권6호
• /
• pp.227-233
• /
• 2019

최근 딥러닝 기술의 발달과 함께 신경 네트워크는 컴퓨터 비전에서도 성공을 거두고 있다. 컨볼루션 신경망은 단순한 영상 분류 작업뿐만 아니라 객체 분할 및 검출 등 난이도가 높은 작업에서도 탁월한 성능을 보였다. 그러나 그러한 많은 심층 학습 모델은 지도학습에 기초하고 있으며, 이는 이미지 라벨보다 주석 라벨이 더 많이 필요하다. 특히 semantic segmentation 모델은 훈련을 위해 픽셀 수준의 주석을 필요로 하는데, 이는 매우 중요하다. 이 논문은 이러한 문제를 해결하기 위한 네트워크 훈련을 위해 영상 수준 라벨만 필요한 약지도 semantic segmentation 방법을 제안한다. 기존의 약지도학습 방법은 대상의 특정 영역만 탐지하는 데 한계가 있다. 반면에, 본 논문에서는 우리의 모델이 사물의 더 다른 부분을 인식하도 multi-classifier 심층 학습 아키텍처를 사용한다. 제안된 방법은 VOC 2012 검증 데이터 세트를 사용하여 평가한다.

• 디지털융복합연구
• /
• 제19권6호
• /
• pp.251-258
• /
• 2021

본 논문에서는 U-Net 기반의 semantic segmentation 방법에서 정확도를 향상시키기 위해 residual learning을 활용한 인코더-디코더 구조의 모델을 제안하였다. U-Net은 딥러닝 기반의 semantic segmentation 방법이며 자율주행 자동차, 의료 영상 분석과 같은 응용 분야에서 주로 사용된다. 기존 U-Net은 인코더의 얕은 구조로 인해 특징 압축 과정에서 손실이 발생한다. 특징 손실은 객체의 클래스 분류에 필요한 context 정보 부족을 초래하고 segmentation 정확도를 감소시키는 문제가 있다. 이를 개선하기 위해 제안하는 방법은 기존 U-Net에 특징 손실과 기울기 소실 문제를 방지하는데 효과적인 residual learning을 활용한 인코더를 통해 context 정보를 효율적으로 추출하였다. 또한, 인코더에서 down-sampling 연산을 줄여 특징맵에 포함된 공간 정보의 손실을 개선하였다. 제안하는 방법은 Cityscapes 데이터셋 실험에서 기존 U-Net 방법에 비해 segmentation 결과가 약 12% 향상되었다.

• Smart Structures and Systems
• /
• 제31권4호
• /
• pp.325-334
• /
• 2023

Computer vision-based damage detection enables non-contact, efficient and low-cost bridge health monitoring, which reduces the need for labor-intensive manual inspection or that for a large number of on-site sensing instruments. By leveraging recent semantic segmentation approaches, we can detect regions of critical structural components and identify damages at pixel level on images. However, existing methods perform poorly when detecting small and thin damages (e.g., cracks); the problem is exacerbated by imbalanced samples. To this end, we incorporate domain knowledge to introduce a hierarchical semantic segmentation framework that imposes a hierarchical semantic relationship between component categories and damage types. For instance, certain types of concrete cracks are only present on bridge columns, and therefore the noncolumn region may be masked out when detecting such damages. In this way, the damage detection model focuses on extracting features from relevant structural components and avoid those from irrelevant regions. We also utilize multi-scale augmentation to preserve contextual information of each image, without losing the ability to handle small and/or thin damages. In addition, our framework employs an importance sampling, where images with rare components are sampled more often, to address sample imbalance. We evaluated our framework on a public synthetic dataset that consists of 2,000 railway bridges. Our framework achieves a 0.836 mean intersection over union (IoU) for structural component segmentation and a 0.483 mean IoU for damage segmentation. Our results have in total 5% and 18% improvements for the structural component segmentation and damage segmentation tasks, respectively, compared to the best-performing baseline model.

• 대한임베디드공학회논문지
• /
• 제16권5호
• /
• pp.209-214
• /
• 2021

Panoptic segmentation, which is now widely used in computer vision such as medical image analysis, and autonomous driving, helps understanding an image with holistic view. It identifies each pixel by assigning a unique class ID, and an instance ID. Specifically, it can classify 'thing' from 'stuff', and provide pixel-wise results of semantic prediction and object detection. As a result, it can solve both semantic segmentation and instance segmentation tasks through a unified single model, producing two different contexts for two segmentation tasks. Semantic segmentation task focuses on how to obtain multi-scale features from large receptive field, without losing low-level features. On the other hand, instance segmentation task focuses on how to separate 'thing' from 'stuff' and how to produce the representation of detected objects. With the advances of both segmentation techniques, several panoptic segmentation models have been proposed. Many researchers try to solve discrepancy problems between results of two segmentation branches that can be caused on the boundary of the object. In this survey paper, we will introduce the concept of panoptic segmentation, categorize the existing method into two representative methods and explain how it is operated on two methods: top-down method and bottom-up method. Then, we will analyze the performance of various methods with experimental results.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제14권11호
• /
• pp.4426-4442
• /
• 2020

Since scene segmentation is becoming a hot topic in the field of autonomous driving and medical image analysis, researchers are actively trying new methods to improve segmentation accuracy. At present, the main issues in image semantic segmentation are intra-class inconsistency and inter-class indistinction. From our analysis, the lack of global information as well as macroscopic discrimination on the object are the two main reasons. In this paper, we propose a Densely connected residual Attention network (DA-Res2Net) which consists of a dense residual network and channel attention guidance module to deal with these problems and improve the accuracy of image segmentation. Specifically, in order to make the extracted features equipped with stronger multi-scale characteristics, a densely connected residual network is proposed as a feature extractor. Furthermore, to improve the representativeness of each channel feature, we design a Channel-Attention-Guide module to make the model focusing on the high-level semantic features and low-level location features simultaneously. Experimental results show that the method achieves significant performance on various datasets. Compared to other state-of-the-art methods, the proposed method reaches the mean IOU accuracy of 83.2% on PASCAL VOC 2012 and 79.7% on Cityscapes dataset, respectively.