Proceedings of the 4th International Conference on Advances in Image Processing最新文献

筛选
英文 中文
Feature Based Deep Retinex for Low-Light Image Enhancement 基于特征的低光图像增强深度视网膜
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441270
Huabao Zhong, Zhang Fan, Liyin Xie, Xu Bin, Junliang Wang, Guijun Liu, Dong Lin
{"title":"Feature Based Deep Retinex for Low-Light Image Enhancement","authors":"Huabao Zhong, Zhang Fan, Liyin Xie, Xu Bin, Junliang Wang, Guijun Liu, Dong Lin","doi":"10.1145/3441250.3441270","DOIUrl":"https://doi.org/10.1145/3441250.3441270","url":null,"abstract":"Low-light image processing is a common issue in industry, media and other practical application fields. Enhancing image brightness or contrast directly may bring accompanying noise and color cast. The proposed method is a feature based deep F-Retinex-Net. A neural network is used to decompose the image into features representation of reflectance component and illumination component. The feature of illumination component was enhanced, and then the features are used to reconstruct the enhanced image by another network. The convergence of the observation model is confirmed by observing the state of illumination and reflectance features by visualization method. Experimental results show that F-Retinex-Net has better performance enhancement effect than Retinex-Net, and can also greatly alleviate the color distortion problem.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126488012","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 1
Fully Convolutional One-Stage Circular Object Detector on Medical Images 医学图像上的全卷积单阶段圆形目标检测器
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441269
Yilong Guo, Fei Chen, Qi Cheng, Jiamin Wu, Bo Wang, Y. Wu, Wenxin Zhao
{"title":"Fully Convolutional One-Stage Circular Object Detector on Medical Images","authors":"Yilong Guo, Fei Chen, Qi Cheng, Jiamin Wu, Bo Wang, Y. Wu, Wenxin Zhao","doi":"10.1145/3441250.3441269","DOIUrl":"https://doi.org/10.1145/3441250.3441269","url":null,"abstract":"In an endoscopic thyroidectomy, parathyroid injury can lead to hypocalcemia for life. It is of great help to assist surgeons to complete endoscopic thyroidectomy without damaging the parathyroid. Considering that the shape of the parathyroid is usually round, we replaced the rectangle bounding box with the circular bounding box, and propose a fully convolutional one-stage circular object detector (FCOSC) based on FCOS. The loss and optimization are improved to adapt to the calculation of the circular bounding box. FCOSC achieved 37.9% in AP with single-model testing on the parathyroid dataset, which was 5% higher than FCOS. Considering that our data is strictly confidential, we also evaluated the accuracy of our method in the BCCD dataset and achieved 70.4% in AP, comparing to 68.4% with a rectangle bounding box. Experiments show that the proposed FCOSC framework can obtain expected results on parathyroid detection and be used in another metical circular object detection task.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"31 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130965907","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 2
3D Object Detection Based on Feature Pyramid Network 基于特征金字塔网络的三维目标检测
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441257
Yang Zhao, Yiguang Liu, Yunan Zheng, Tianhang Tang
{"title":"3D Object Detection Based on Feature Pyramid Network","authors":"Yang Zhao, Yiguang Liu, Yunan Zheng, Tianhang Tang","doi":"10.1145/3441250.3441257","DOIUrl":"https://doi.org/10.1145/3441250.3441257","url":null,"abstract":"3D object detection aims to study how to perceive environmental information effectively, classify and locate interested objects accurately. In order to solve the problem that the object is easy to be lost in complex environments (such as partial occlusion, at long distance), we proposed a multi-scale 3D object detection method based on feature pyramid network (FPN). Firstly, input RGB image corresponding to the scene and bird's eye view (BEV) map into the pyramid feature extractor to construct multi-scale and strong semantic feature representations. Secondly, priori anchor boxes are applied to each layer of the feature pyramid, and the regional fusion features are obtained and input to the classifier and regressor. Finally, cross-scale detection is completed to obtain the best classification, dimensions and orientation estimation results. The experiments are conducted on the KITTI dataset and the results show that FPN-based feature extraction and cross-scale detection method can effectively improve the detection ability of partial occluded and long-distance objects.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133373286","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Automatic Evaluation of Difficult-to-Recognize Characters for Efficient Transcription of Japanese Cursive Characters 日语草书高效转录难识别字符的自动评价
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441255
K. Ueki, Tomoka Kojima
{"title":"Automatic Evaluation of Difficult-to-Recognize Characters for Efficient Transcription of Japanese Cursive Characters","authors":"K. Ueki, Tomoka Kojima","doi":"10.1145/3441250.3441255","DOIUrl":"https://doi.org/10.1145/3441250.3441255","url":null,"abstract":"In this study, we approached Japanese cursive character (kuzushiji) recognition using machine learning to improve the efficiency of their transcription process. In particular, we discussed methods that automatically distinguish difficult-to-recognize characters or characters that were not included in the training, and put them on hold as unknown characters. We analyzed two methods that deferred recognition of characters: In the first method, characters with low confidence values obtained from the recognition model were put on hold as unknown characters. In the second method, we trained a “miscellaneous class” that combined multiple character classes and deferred the recognition of characters if they were recognized as belonging to the “miscellaneous class.” Experiments showed that by narrowing down the characters to be recognized, a recognition rate of more than 90% was achieved for over 1,500 character classes, confirming the usefulness of temporarily deferring recognition during the transcription process.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114938677","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
A Novel Algorithm of Real-time Aerial Image Mosaic 一种新的实时航拍图像拼接算法
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441252
Yonghui Xi, Shiqiang Hu
{"title":"A Novel Algorithm of Real-time Aerial Image Mosaic","authors":"Yonghui Xi, Shiqiang Hu","doi":"10.1145/3441250.3441252","DOIUrl":"https://doi.org/10.1145/3441250.3441252","url":null,"abstract":"Aerial image mosaic is to stitch the images taken by UAV (Unmanned aerial vehicles) from different positions into a panoramic image, which has attracted significant interest in enemy reconnaissance, terrain detection and environmental monitoring. Aiming at the problem of low process speed in the traditional algorithm, this paper proposes a real-time aerial image mosaic algorithm. First, we take advantage of ORB feature for feature extraction, and then KNN (K-Nearest Neighbor) algorithm based on Hash table is applied for fast feature matching. In order to make the algorithm more robust, we use RANSAC (Random Sampling Consistency) method to remove mismatched feature points in the process of estimating the transformation matrix. In the stage of image registration, we propose a new coordinates unification method based on the background image, which greatly reduces the amount of computation. Finally, we present a method to find the overlapped areas of images effectively and the distance weighted fusion algorithm is used in the overlapped areas to eliminate seams between images. Experiments show that the image mosaic algorithm proposed in this paper has both real-time performance and good stitching quality. Additional Keywords and Phrases: Feature matching, Image registration, Image fusion","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128948413","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
Object Detection with Auto-Learning Anchor Algorithm 基于自动学习锚算法的目标检测
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441262
Tingsong Ma, Wenhong Tian, Yuanlun Xie
{"title":"Object Detection with Auto-Learning Anchor Algorithm","authors":"Tingsong Ma, Wenhong Tian, Yuanlun Xie","doi":"10.1145/3441250.3441262","DOIUrl":"https://doi.org/10.1145/3441250.3441262","url":null,"abstract":"As an effective auxiliary means for object detection task, region anchors are widely adopted in most of state-of-the art detectors. However, anchor's location and shape in those works are normally determined by experience or some preprocessing methods, i.e., clustering, which leads to time consumption and limits the flexibility of anchors. In this paper, we explore the possibility that networks can predict bounding boxes and simultaneously learn anchor's location and shape end to end. Specifically, we propose a new anchoring scheme, named Automatic Anchor Learning, which can be integrated into any object detectors and enable detectors to learn the location and size of anchors while training, without sampling anchors over any predefined set of scales and aspect ratios. The proposed method first predicts where the centers of objects of interest might exist and then predict the shape of anchor that should be placed in this location. By applying the proposed Automatic Anchoring Learning method to Yolov3 model, we achieve around 3.3% and 1.6% higher recall and mAP on MS COCO with 80% less anchors, and 10% more FPS than the original Yolov3. Additionally, we also integrate our method into other object algorithms, i.e., Fast R-CNN and RetinaNet, we respectively improve their detection mAP by 2.5% and 1.1%.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116033205","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 2
Horse Breed Classification Based on Transfer Learning 基于迁移学习的马品种分类
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441264
Yang Fu, Xiangnian Huang, Yun Li
{"title":"Horse Breed Classification Based on Transfer Learning","authors":"Yang Fu, Xiangnian Huang, Yun Li","doi":"10.1145/3441250.3441264","DOIUrl":"https://doi.org/10.1145/3441250.3441264","url":null,"abstract":"Expert identification of horse breeds is an age-old task that can now be identified using genetic techniques. However, neither approach is cheap nor efficient. The automatic classification of horse breeds by computer vision is an effective solution. In this paper, we solve this task by proposing a novel method using transfer learning of pre-trained deep convolution neural networks architectures. The pre-trained convolutional neural networks include MobilenetV2, Mobilenet, Xception, VGG16, and VGG19. We use the keras deep learning framework, and train these deep convolution neural networks for transfer learning, which overcomes the problem of small amount of data in the early stage. An extensive experimental study on various horse breeds datasets shows that our method obtains an average accuracy rate of automatic classification of horse breeds to 89.34%, which has obvious advantage over other deep convolutional neural network models such as xception, vgg16, vgg19 and self-made convolutional neural network.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130182341","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 2
Multi-Dimensional Spatial Attention Residual U-Net (Msaru-Net) for Low-Dose Lung Ct Image Restoration 多维空间注意残差U-Net (Msaru-Net)用于低剂量肺Ct图像恢复
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441266
Zhiwei Feng, Yizhong Wang, Ailong Cai
{"title":"Multi-Dimensional Spatial Attention Residual U-Net (Msaru-Net) for Low-Dose Lung Ct Image Restoration","authors":"Zhiwei Feng, Yizhong Wang, Ailong Cai","doi":"10.1145/3441250.3441266","DOIUrl":"https://doi.org/10.1145/3441250.3441266","url":null,"abstract":"Low-dose computed tomography (LDCT) is considered as the most effective method for early lung cancer screening in the clinic. However, it's still a significant challenge to ensure the quality of reconstructed images under the premise of reducing the radiation dose. To reduce the interference of image quality degradation on the diagnosis of radiologists, in this paper, we introduce a multi-dimensional spatial attention residual U-net (MSARU-net) model to improve the low-dose lung CT images quality. The proposed MSARU-net combines residual U-net, multi-dimensional feature extraction module and spatial attention mechanism to retain critical structural information of different levels while suppressing image noise. Experiments with simulation LDCT data have shown its superiority in visual effects and quantitative measurements, indicating the potential of achieving a noticeable gain compared to the state-of-the-art methods.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"43 3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126989049","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 1
An Improved Particle Swarm Optimization Algorithm for Pigeon Brain Tissue MRI Image Segmentation 一种改进的粒子群算法用于鸽子脑组织MRI图像分割
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441256
Liwei Chen, Lina Zhao, Deqing Yang
{"title":"An Improved Particle Swarm Optimization Algorithm for Pigeon Brain Tissue MRI Image Segmentation","authors":"Liwei Chen, Lina Zhao, Deqing Yang","doi":"10.1145/3441250.3441256","DOIUrl":"https://doi.org/10.1145/3441250.3441256","url":null,"abstract":"Pigeons have strong disease resistance and short reproductive cycle. It is an ideal sample for bird research. It has been widely used in scientific research and related experiments in the fields of biology, neurology, and medicine. Generally, after obtaining pigeon brain slices or images, a segmentation algorithm is used to extract brain tissue for further research. The segmentation accuracy of the algorithm has always been the focus and difficulty in the research. This paper used Otsu algorithm, fuzzy threshold segmentation algorithm, K-means algorithm and improved particle swarm algorithm to segment pigeon brain tissue. We calibrated the area of pigeon brain tissue by comparing the segmented image with the labeled pigeon brain tissue. According to certain evaluation criteria, the accuracy of brain tissue segmentation in different MRI images was calculated. By comparing the accuracy of each tissue segmentation, we can quantitatively judge the segmentation effect of the algorithm. Comprehensive analysis showed that the improved particle swarm optimization algorithm had a better segmentation effect.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130336434","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
The Multi-Watermarks Attack of DNN Watermarking DNN水印的多水印攻击
Proceedings of the 4th International Conference on Advances in Image Processing Pub Date : 2020-11-13 DOI: 10.1145/3441250.3441279
Deyin Li, Yu Yang
{"title":"The Multi-Watermarks Attack of DNN Watermarking","authors":"Deyin Li, Yu Yang","doi":"10.1145/3441250.3441279","DOIUrl":"https://doi.org/10.1145/3441250.3441279","url":null,"abstract":"Deep learning models are widely used in business scenarios and have achieved some success. It is usually time or computing consuming to build a production-level deep learning model. As a result, such models require copyright protection by watermarks. So the security of watermarks is important. In this paper, multi-watermarks attack, which can prevent the model owner from declaring his ownership of the model, is proposed. In special cases, it can completely remove the watermarks which are based on the output of models and it can decrease the watermark accuracy down to less than 15% with only 5 rounds of retraining. Besides, it can also be used to remove the backdoor in models or decrease the task accuracy of models.","PeriodicalId":205484,"journal":{"name":"Proceedings of the 4th International Conference on Advances in Image Processing","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-11-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132546170","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
引用次数: 0
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
相关产品
×
本文献相关产品
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信
小红书