- 标题
- 摘要
- 关键词
- 实验方案
- 产品
-
[IEEE 2019 IEEE International Conference on Image Processing (ICIP) - Taipei, Taiwan (2019.9.22-2019.9.25)] 2019 IEEE International Conference on Image Processing (ICIP) - Prior Knowledge Guided Small Object Detection on High-Resolution Images
摘要： When applying common object detection algorithms to detect small objects on high-resolution images, the down-sampling operation of the input images is inevitable due to the limitation of GPU memory. Accordingly, the details for characterizing small objects are lost. To resolve this contradiction, a small object detection method in a coarse-to-fine manner is presented. Specifically, some rough regions of interest (ROI) are firstly computed from low-resolution images. The prior knowledge of the positions of objects is used to guide the generation of ROIs. Then the features of small ROIs are recomputed from high-resolution images, and the features of large ROIs are obtained from the feature maps used to generate ROIs. The proposed method is validated on two datasets. One is a plant phenotyping dataset and the other is a public traffic sign dataset. Experimental results convincingly show the effectiveness of the proposed method.
关键词： prior knowledge,convolutional neural network,high-resolution image,small object detection
更新于2025-09-11 14:15:04
-
[IEEE 2018 International Symposium ELMAR - Zadar, Croatia (2018.9.16-2018.9.19)] 2018 International Symposium ELMAR - Image Feature Matching and Object Detection Using Brute-Force Matchers
摘要： The paper considers a problem of feature matching and object detection in two images using brute-force matchers. The proposed framework exploited several concurrent algorithms for feature detection and descriptor extraction, such as ORB (Oriented FAST and Rotated BRIEF), BRISK (Binary Robust Invariant Scalable Keypoints), SIFT (Scale Invariant Feature Transform) and SURF (Speeded-Up Robust Features). The feature matching is accomplished by the Brute-Force approach combined with the k-Nearest Neighbors algorithm. The obtained matches are utilized by the robust RANSAC (Random Sample Consensus) method for estimating the transformation between two consecutive images. Therefore, the RANSAC method is employed to improve the outliers removal. The proposed algorithm is designed and implemented using OpenCV library. Its effectiveness and quality are veri?ed through analyses of its execution speed and accuracy of the feature matching.
关键词： Feature matching,Brute-force algorithm,Object detection,RANSAC,Feature detection and extraction
更新于2025-09-10 09:29:36
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Object Detection with Head Direction in Remote Sensing Images Based on Rotational Region CNN
摘要： Object detection has been playing a significant role in the field of remote sensing for a long time but it is still full of challenges. In this paper, we propose a novel detection framework based on rotational region convolution neural network to cope with the problem of non-maximum suppression in dense objects detection. The bounding boxes obtained by adopting our method is the minimum bounding rectangle of object with less redundant regions. Furthermore, we find the head direction of the object through prediction. There are three important changes to our framework over traditional detection methods, representation and regression of rotational bounding box, head direction prediction and rotational non-maximal suppression. Experiments based on remote sensing images from Google Earth for Object detection show that our detection method based on rotational region CNN has a competitive performance.
关键词： prediction,object detection,rotating region,convolution neural network,non-maximal suppression
更新于2025-09-10 09:29:36
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Semi-Supervised Object Detection in Remote Sensing Images Using Generative Adversarial Networks
摘要： Object detection is a challenging task in computer vision. Now many detection networks can get a good detection result when applying large training dataset. However, annotating sufficient amount of data for training is often time-consuming. To address this problem, a semi-supervised learning based method is proposed in this paper. Semi-supervised learning trains detection networks with few annotated data and massive amount of unannotated data. In the proposed method, Generative Adversarial Network is applied to extract data distribution from unannotated data. The extracted information is then applied to improve the performance of detection network. Experiment shows that the method in this paper greatly improves the detection performance compared w1ith supervised learning using only few annotated data. The results prove that it is possible to achieve acceptable detection result when only few target object is annotated in the training dataset.
关键词： generative adversarial networks (GAN),convolutional neural networks (CNN),Semi-supervised learning,object detection
更新于2025-09-10 09:29:36
-
[IEEE 2018 International Conference on Network Infrastructure and Digital Content (IC-NIDC) - Guiyang, China (2018.8.22-2018.8.24)] 2018 International Conference on Network Infrastructure and Digital Content (IC-NIDC) - Ship detection in foggy remote sensing image via scene classification R-CNN
摘要： The object detection networks via Faster R-CNN for ship detection have demonstrated impressive performance. However, the complexity of weather conditions in high resolution satellite images exposes the limited capacity of these networks. Images interfered by fog are common in optical remote sensing images. In this paper, we embrace this observation and introduce our research. Unlike SAR images, optical sensor images are very susceptible to the effects of the weather, especially clouds and fog.So, accurate target information cannot be obtained from these image, which reduces the accuracy of ship detection. To solve this problem, we attempts to introduce the image defogging methods into object detection networks to suppress the interference of clouds. Secondly, the SC-R-CNN structure is proposed, which uses the scene classification network (SCN) to realize the classification of fog-containing images and cascaded with the object detection network to form a dual-stream object detection framework. In addition, the combination of defogging methods and the SC-R-CNN network also produces more optimized results. We use the remote sensing image data set containing various types of weather conditions to confirm the validity and accuracy of the proposed method.
关键词： Remote sensing,Image processing,Defogging,Object detection,Convolutional neural network,Deep learning
更新于2025-09-10 09:29:36
-
Real-Time Grayscale Dehazing Scheme For Car Vision
摘要： To improve the safety of autonomous cars, their obstacle detection capability in bad weather must be substantially improved. Haze is a major factor that degrades outdoor images. Although various dehazing schemes have been proposed, a dehazing scheme designed to improve obstacle detection capability has not been reported. Hence, we present a dehazing algorithm that enhances the safety of an autonomous car. This algorithm should be able to work in real time, even using edge computers typically installed as car electronics. Furthermore, this algorithm should work on grayscale images, as systems dependent on color images are often unaffected by environmental color changes caused by factors such as a setting sun. We developed this algorithm based on the following three existing dehazing algorithms: dark channel prior, median dark channel prior, and the parameter tuning scheme for dark channel prior. We extend these methods based only on grayscale images. In terms of object detection capability, structural similarity index measure, and peak signal-to-noise ratio, the empirical results showed that our grayscale image-based proposed algorithm is comparable to the results of current cutting-edge methods, and operates in real time.
关键词： Object detection,Autonomous cars,Dehazing
更新于2025-09-10 09:29:36
-
[IEEE IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Valencia, Spain (2018.7.22-2018.7.27)] IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium - Salient Object Detection Via Double Sparse Representations Under Visual Attention Guidance
摘要： This paper introduces a novel method for salient object detection from the perspective of sparse representation under visual attention guidance. After pretreatment and regional analysis with eye fixation detection and multi scale segmentation, regions that are used to make up the foreground and background dictionaries are respectively selected by sorting the visual attraction level of all image regions. For saliency measurement, the reconstruction errors instead of common local and global contrasts are used as the saliency indicator, which is expected to improve the object integrity. In addition, the multi scale workflow is conductive to enhance the robustness for objects of different sizes. The proposed method was compared to six state-of-the-art saliency detection methods using three benchmark datasets, and it was confirmed to have more favorable performance in the detection of multiple objects as well as maintaining the integrity of the object area.
关键词： Salient object detection,visual attention guidance,reconstruction error,sparse representation
更新于2025-09-10 09:29:36
-
[IEEE 2017 International Conference on Progress in Informatics and Computing (PIC) - Nanjing (2017.12.15-2017.12.17)] 2017 International Conference on Progress in Informatics and Computing (PIC) - Polarization and solar altitude correlation analysis and application in object detection
摘要： Complicated optical scenes generally exist on the water surface. The noise originated from the light reflection seriously blocks the performance of the object detection method. This paper analyzes the correlation between the incident light angle and the flare noise, the relation between the solar altitude and the polarization state of the reflected light. Further, the correlation between the various imaging factors, such as the imaging time, the imaging angle of view, the polarization direction and the polarization state, is adjusted and optimized by the polarization imaging experiment. According to the experimental results, our polarization imaging method has good capability to suppress the noises generated by the light reflection, improving the accuracy of the object detection results.
关键词： Object Detection,Solar Altitude,Polarization Analysis,Polarization Imaging
更新于2025-09-10 09:29:36
-
[IEEE 2018 International Conference on Content-Based Multimedia Indexing (CBMI) - La Rochelle (2018.9.4-2018.9.6)] 2018 International Conference on Content-Based Multimedia Indexing (CBMI) - Active Learning to Assist Annotation of Aerial Images in Environmental Surveys
摘要： Nowadays, remote sensing technologies greatly ease environmental assessment using aerial images. Such data are most often analyzed by a manual operator, leading to costly and non scalable solutions. In the fields of both machine learning and image processing, many algorithms have been developed to fasten and automate this complex task. Their main common assumption is the need to have prior ground truth available. However, for field experts or engineers, manually labeling the objects requires a time-consuming and tedious process. Restating the labeling issue as a binary classification one, we propose a method to assist the costly annotation task by introducing an active learning process, considering a query-by-group strategy. Assuming that a comprehensive context may be required to assist the annotator with the labeling task of a single instance, the labels of all the instances of an image are indeed queried. A score based on instances distribution is defined to rank the images for annotation and an appropriate retraining step is derived to simultaneously reduce the interaction cost and improve the classifier performances at each iteration. A numerical study on real images is conducted to assess the algorithm performances. It highlights promising results regarding the classification rate along with the chosen re-training strategy and the number of interactions with the user.
关键词： human-in-the-loop,aerial images,object detection,Active learning,data annotation
更新于2025-09-10 09:29:36
-
[Lecture Notes in Electrical Engineering] Advanced Multimedia and Ubiquitous Engineering Volume 518 (MUE/FutureTech 2018) || Superpixel Based ImageCut Using Object Detection
摘要： The edge preserving image segmentation required by online shopping malls or the design ?eld is clearly limited to pixel based image machine learning, making it dif?cult for the industry to accept the results of the latest machine learning techniques. Existing studies of image segmentation have shown that using any size square as a study unit without targeting meaningful pixels provides a simple method of learning, but produces a high error rate in image segmentation and also there is no way to calibrate the resulting images. Therefore, this paper proposes image segmentation techniques through superpixel based machine learning to develop technologies for automatically identifying and separating objects from images. In addition, the main reasons for superpixel based imagecut using object detection is to reduce the amount of data processed, thereby effectively delivering higher computational rates and larger image processing.
关键词： Removal image background,Superpixel,Image segmentation,Machine learning,Object detection
更新于2025-09-10 09:29:36