Regions With Convolutional Neural Network Research Articles

Aircraft Detection System Based on Regions with Convolutional Neural Networks

Object detection in remote sensing imagery is an important topic of image processing researches. Detection of objects and regions from satellite imagery takes its place in various applications such as the detection of residential areas and agricultural lands, road lines, ships, as well as airport and hangar detections. As a more specific remote sensing imagery based object-detection process, a stationary aircraft detection system could serve as a model in some military applications. Such a model could serve the detection of stationary aircraft targets in airports. In the proposed study, a deep learning-based model detects the aircraft in the airports using the satellite images from Google Earth. The deep learning model uses the state of the art Regions with Convolutional Neural Network (RCNN). Firstly a built from scratch CNN design is used for the basic learning step of the system. Then, RCNN performs region detection, which anchors the stationary aircraft object bounding boxes. A large dataset containing aircraft images is preferred for the training of CNN. To validate the system, satellite images captured from airports in Turkey are used. The results of the study show that the proposed model successfully operates aircraft detection with high-performance rates. While the classifier network structure, which constitutes the first step of the study, produces 98.4% test accuracy, the proposed aircraft detection framework has successfully performed the aircraft identification process by producing matched bounding boxes in the test images.

Automatic Segmentation of Meniscus in Multispectral MRI Using Regions with Convolutional Neural Network (R-CNN).

The meniscus has a significant function in human anatomy, and Magnetic Resonance Imaging (MRI) has an essential role in meniscus examination. Due to a variety of MRI data, it is excessively difficult to segment the meniscus with image processing methods. An MRI data sequence contains multiple images, and the region features we are looking for may vary from each image in the sequence. Therefore, feature extraction becomes more difficult, and hence, explicitly programming for segmentation becomes more difficult. Convolutional Neural Network (CNN) extracts features directly from images and thus eliminates the need for manual feature extraction. Regions with Convolutional Neural Network (R-CNN) allow us to use CNN features in object detection problems by combining CNN features with Region Proposals. In this study, we designed and trained an R-CNN for detecting meniscus region in MRI data sequence. We used transfer learning for training R-CNN with a small amount of meniscus data. After detection of the meniscus region by R-CNN, we segmented meniscus by morphological image analysis using two different MRI sequences. Automatic detection of the meniscus region with R-CNN made the meniscus segmentation process easier, and the use of different contrast features of two different image sequences allowed us to differentiate the meniscus from its surroundings.

Classification and detection of chili and its flower using deep learning approach

This study presents an application of using Deep Neural Network (DNN) based detector to detect chili and its flower in the chili plant image. Detecting chili on its plant is important for the development of a robotic vision to automatically picking the chili. Only one type of local chili variety is used in this study from the species of Capsicum frustecens. Five hundred of chili plant images were captured from multiple angles and each of images was marked and labelled for any present of chili and its flower. These images were divided into 70-30 per cent proportion for training with validation and testing purposes accordingly. This project uses Faster Regions with Convolutional Neural Networks (R-CNN) as a deep learning model for training that contains around 177 numbers of layers including input and output layers. The classifier network was trained to optimize all parameters involved in chili and its flower classification and detection. The classification and detection accuracy are measured on the tested images. The result shows very good accuracy in validation and testing for classification and detection especially with the image of chili plan is upright position.

Stereo Vision-Based Object Recognition and Manipulation by Regions with Convolutional Neural Network

This paper develops a hybrid algorithm of adaptive network-based fuzzy inference system (ANFIS) and regions with convolutional neural network (R-CNN) for stereo vision-based object recognition and manipulation. The stereo camera at an eye-to-hand configuration firstly captures the image of the target object. Then, the shape, features, and centroid of the object are estimated. Similar pixels are segmented by the image segmentation method, and similar regions are merged through selective search. The eye-to-hand calibration is based on ANFIS to reduce computing burden. A six-degree-of-freedom (6-DOF) robot arm with a gripper will conduct experiments to demonstrate the effectiveness of the proposed system.

Information-Guided Flame Detection Based on Faster R-CNN

Due to the diversity of the shape and texture of flame, and interference objects that similar to flame in color, detecting the position of flame from images is a difficult task. To enable generic object detection methods to achieve better performance in flame detection tasks, a color-guided anchoring strategy is proposed that uses color features of the flame to limit the location of the anchor. To solve the problem of high false alarm rate when directly using generic object detection methods in flame detection, a global information-guided flame detection method is proposed, this strategy uses a parallel network to generate image global information. We use these two methods to improve Faster R-CNN (Regions with Convolutional Neural Network features) to perform the fire detection process in a guided manner. Experiments on the BoWFire dataset show that our method improved detection speed by 10.1% compared with the original Faster R-CNN. In addition, the false alarm rate is decreased by 21.5%, and the overall accuracy of detection is increased by 9.3%. Experiments on PascalVOC and Corsician datasets further demonstrate the robustness of the proposed methods.

Two-stage framework for optic disc localization and glaucoma classification in retinal fundus images using deep learning

BackgroundWith the advancement of powerful image processing and machine learning techniques, Computer Aided Diagnosis has become ever more prevalent in all fields of medicine including ophthalmology. These methods continue to provide reliable and standardized large scale screening of various image modalities to assist clinicians in identifying diseases. Since optic disc is the most important part of retinal fundus image for glaucoma detection, this paper proposes a two-stage framework that first detects and localizes optic disc and then classifies it into healthy or glaucomatous.MethodsThe first stage is based on Regions with Convolutional Neural Network (RCNN) and is responsible for localizing and extracting optic disc from a retinal fundus image while the second stage uses Deep Convolutional Neural Network to classify the extracted disc into healthy or glaucomatous. Unfortunately, none of the publicly available retinal fundus image datasets provides any bounding box ground truth required for disc localization. Therefore, in addition to the proposed solution, we also developed a rule-based semi-automatic ground truth generation method that provides necessary annotations for training RCNN based model for automated disc localization.ResultsThe proposed method is evaluated on seven publicly available datasets for disc localization and on ORIGA dataset, which is the largest publicly available dataset with healthy and glaucoma labels, for glaucoma classification. The results of automatic localization mark new state-of-the-art on six datasets with accuracy reaching 100% on four of them. For glaucoma classification we achieved Area Under the Receiver Operating Characteristic Curve equal to 0.874 which is 2.7% relative improvement over the state-of-the-art results previously obtained for classification on ORIGA dataset.ConclusionOnce trained on carefully annotated data, Deep Learning based methods for optic disc detection and localization are not only robust, accurate and fully automated but also eliminates the need for dataset-dependent heuristic algorithms. Our empirical evaluation of glaucoma classification on ORIGA reveals that reporting only Area Under the Curve, for datasets with class imbalance and without pre-defined train and test splits, does not portray true picture of the classifier’s performance and calls for additional performance metrics to substantiate the results.

Multispectral Heterogeneity Detection Based on Frame Accumulation and Deep Learning

Hyperspectral transmission imaging may provide a means for rapid screening of breast tumors, but tissue has a strong nature of scattering, thus causing a great difficulty in identifying heterogeneity. In this paper, a combination of frame accumulation and deep learning was proposed to detect heterogeneity, and we designed the simulation experiment of collecting phantom images. On the basis of frame accumulation preprocessing, the heterogeneous detection is performed on multispectral images by using faster regions with convolutional neural networks (R-CNN) features and a single shot multibox detector (SSD), two typical detection frameworks of deep learning. The results show that the mean average precision (mAP) of faster R-CNN and SSD reach 90.8% and 95.1%, respectively, when three classes (including background) are detected with the help of the dataset provided in this paper, and the mAP of two frameworks both reach 99.9% when two classes (including background) are detected. The detection efficiency of the SSD is higher than faster, SSD’s detection speed can reach 50 fps, and the detection accuracy of the images after frame accumulation preprocessing is higher than that without frame accumulation processing. In summary, we validate the possibility of employing faster R-CNN and SSD to detect heterogeneity in multispectral images based on frame accumulation that improves image grayscale resolution, and it has a certain degree of reference significance for the application of deep learning in multispectral image detection.

Deep feature based contextual model for object detection

Object detection is one of the most active areas in computer vision, which has made significant improvement in recent years. Current state-of-the-art object detection methods mostly adhere to the framework of regions with convolutional neural network (R-CNN) and only use local appearance features inside object bounding boxes. Since these approaches ignore the contextual information around the object proposals, the outcome of these detectors may generate a semantically incoherent interpretation of the input image. In this paper, we propose an ensemble object detection system which incorporates the local appearance, the contextual information in term of relationships among objects and the global scene based contextual feature generated by a convolutional neural network. The system is formulated as a fully connected conditional random field (CRF) defined on object proposals and the contextual constraints among object proposals are modeled as edges naturally. Furthermore, a fast mean field approximation method is utilized to inference in this CRF model efficiently. The experimental results demonstrate that our approach achieves a higher mean average precision (mAP) on PASCAL VOC 2007 datasets compared to the baseline algorithm Faster R-CNN.

Facial Expression Recognition with Faster R-CNN

In order to avoid the complex explicit feature extraction process and the problem of low-level data operation involved in traditional facial expression recognition, we proposed a method of Faster R-CNN (Faster Regions with Convolutional Neural Network Features) for facial expression recognition in this paper. Firstly, the facial expression image is normalized and the implicit features are extracted by using the trainable convolution kernel. Then, the maximum pooling is used to reduce the dimensions of the extracted implicit features. After that, RPNs (Region Proposal Networks) is used to generate high-quality region proposals, which are used by Faster R-CNN for detection. Finally, the Softmax classifier and regression layer is used to classify the facial expressions and predict boundary box of the test sample, respectively. The dataset is provided by Chinese Linguistic Data Consortium (CLDC), which is composed of multimodal emotional audio and video data. Experimental results show the performance and the generalization ability of the Faster R-CNN for facial expression recognition. The value of the mAP is around 0.82.