Enhanced multi-scale networks for semantic segmentation

分割 计算机科学 人工智能 像素 模式识别(心理学) 特征(语言学) 相似性(几何) 背景(考古学) 尺度空间分割 频道(广播) 比例(比率) 图像分割 图像(数学) 古生物学 计算机网络 哲学 语言学 物理 量子力学 生物
作者
Tianping Li,Zhaotong Cui,Han Yu,Guanxing Li,Meng Li,Dongmei Wei
出处
期刊:Complex & Intelligent Systems 卷期号:10 (2): 2557-2568 被引量:4
标识
DOI:10.1007/s40747-023-01279-x
摘要

Abstract Multi-scale representation provides an effective answer to the scale variation of objects and entities in semantic segmentation. The ability to capture long-range pixel dependency facilitates semantic segmentation. In addition, semantic segmentation necessitates the effective use of pixel-to-pixel similarity in the channel direction to enhance pixel areas. By reviewing the characteristics of earlier successful segmentation models, we discover a number of crucial elements that enhance segmentation model performance, including a robust encoder structure, multi-scale interactions, attention mechanisms, and a robust decoder structure. The attention mechanism of the asymmetric non-local neural network (ANNet) is merged with multi-scale pyramidal modules to accelerate model segmentation while maintaining high accuracy. However, ANNet does not account for the similarity between pixels in the feature map channel direction, making the segmentation accuracy unsatisfactory. As a result, we propose EMSNet, a straightforward convolutional network architecture for semantic segmentation that consists of Integration of enhanced regional module (IERM) and Multi-scale convolution module (MSCM). The IERM module generates weights using four or five-stage feature maps, then fuses the input features with the weights and uses more computation. The similarity of the channel direction feature graphs is also calculated using ANNet’s auxiliary loss function. The MSCM module can more accurately describe the interactions between various channels, capture the interdependencies between feature pixels, and capture the multi-scale context. Experiments prove that we perform well in tests using the benchmark dataset. On Cityscapes test data, we get 82.2% segmentation accuracy. The mIoU in the ADE20k and Pascal VOC datasets are, respectively, 45.58% and 85.46%.
最长约 10秒,即可获得该文献文件

科研通智能强力驱动
Strongly Powered by AbleSci AI
更新
大幅提高文件上传限制,最高150M (2024-4-1)

科研通是完全免费的文献互助平台,具备全网最快的应助速度,最高的求助完成率。 对每一个文献求助,科研通都将尽心尽力,给求助人一个满意的交代。
实时播报
刚刚
沉静的蜗牛完成签到,获得积分10
刚刚
弈yx发布了新的文献求助10
2秒前
Bob完成签到,获得积分10
2秒前
2秒前
深情安青应助小沈小沈采纳,获得10
2秒前
科研小白完成签到 ,获得积分10
2秒前
3秒前
朱子完成签到,获得积分10
3秒前
3秒前
FashionBoy应助Ali采纳,获得10
4秒前
11发布了新的文献求助30
5秒前
爆米花应助Tree采纳,获得10
6秒前
单薄沐夏完成签到 ,获得积分10
7秒前
慕青应助啤酒白菜采纳,获得10
7秒前
ShawnWei完成签到,获得积分10
8秒前
诸天蓉完成签到,获得积分10
8秒前
李健应助deanna采纳,获得10
8秒前
全或无发布了新的文献求助10
9秒前
9秒前
小米发布了新的文献求助10
9秒前
11秒前
斯文败类应助纪元龙采纳,获得10
11秒前
宝宝熊的熊宝宝完成签到,获得积分10
11秒前
隐形曼青应助念心采纳,获得10
11秒前
lynn完成签到 ,获得积分10
13秒前
Orange应助phraly采纳,获得10
14秒前
ZL完成签到,获得积分10
14秒前
14秒前
Hevesy完成签到,获得积分10
15秒前
16秒前
充电宝应助小米采纳,获得10
16秒前
17秒前
pumpkin发布了新的文献求助10
19秒前
19秒前
火火完成签到 ,获得积分10
19秒前
打打应助精明晓刚采纳,获得10
19秒前
pjh发布了新的文献求助10
20秒前
斯文败类应助大麦迪采纳,获得10
21秒前
shuangcheng发布了新的文献求助10
21秒前
高分求助中
Sustainability in Tides Chemistry 2800
The Young builders of New china : the visit of the delegation of the WFDY to the Chinese People's Republic 1000
юрские динозавры восточного забайкалья 800
English Wealden Fossils 700
Foreign Policy of the French Second Empire: A Bibliography 500
Chen Hansheng: China’s Last Romantic Revolutionary 500
XAFS for Everyone 500
热门求助领域 (近24小时)
化学 医学 生物 材料科学 工程类 有机化学 生物化学 物理 内科学 纳米技术 计算机科学 化学工程 复合材料 基因 遗传学 催化作用 物理化学 免疫学 量子力学 细胞生物学
热门帖子
关注 科研通微信公众号,转发送积分 3143796
求助须知:如何正确求助?哪些是违规求助? 2795335
关于积分的说明 7814709
捐赠科研通 2451390
什么是DOI,文献DOI怎么找? 1304463
科研通“疑难数据库(出版商)”最低求助积分说明 627230
版权声明 601419