Citrus Huanglongbing Detection Based on Multi-Modal Feature Fusion Learning

Citrus Huanglongbing (HLB), also named citrus greening disease, occurs worldwide and is known as a citrus cancer without an effective treatment. The symptoms of HLB are similar to those of nutritional deficiency or other disease. The methods based on single-source information, such as RGB images or hyperspectral data, are not able to achieve great detection performance. In this study, a multi-modal feature fusion network, combining a RGB image network and hyperspectral band extraction network, was proposed to recognize HLB from four categories (HLB, suspected HLB, Zn-deficient, and healthy). Three contributions including a dimension-reduction scheme for hyperspectral data based on a soft attention mechanism, a feature fusion proposal based on a bilinear fusion method, and auxiliary classifiers to extract more useful information are introduced in this manuscript. The multi-modal feature fusion network can effectively classify the above four types of citrus leaves and is better than single-modal classifiers. In experiments, the highest accuracy of multi-modal network recognition was 97.89% when the amount of data was not very abundant (1,325 images of the four aforementioned types and 1,325 pieces of hyperspectral data), while the single-modal network with RGB images only achieved 87.98% recognition and the single-modal network using hyperspectral information only 89%. Results show that the proposed multi-modal network implementing the concept of multi-source information fusion provides a better way to detect citrus HLB and citrus deficiency.

Image Network Teaching Resource Retrieval Algorithm Based on Deep Hash Algorithm

In order to improve the retrieval ability of multiview attribute coded image network teaching resources, a retrieval algorithm of image network teaching resources based on depth hash algorithm is proposed. The pixel big data detection model of the multiview attribute coding image network teaching resources is constructed, the pixel information collected by the multiview attribute coding image network teaching resources is reconstructed, the fuzzy information feature components of the multiview attribute coding image are extracted, and the edge contour distribution image is combined. The distributed fusion result of the edge contour of the view image of the network teaching resources realizes the construction of the view feature parameter set. The gray moment invariant feature analysis method is used to realize information coding, the depth hash algorithm is used to realize the retrieval of multiview attribute coded image network teaching resources, and the information recombination is realized according to the hash coding result of multiview attribute coded image network teaching resources, thus improving the fusion. The simulation results show that this method has higher precision, better retrieval precision, and higher level of resource fusion for multiview coded image network teaching resource retrieval.

Brain2Pix: Fully convolutional naturalistic video reconstruction from brain activity

Reconstructing complex and dynamic visual perception from brain activity remains a major challenge in machine learning applications to neuroscience. Here we present a new method for reconstructing naturalistic images and videos from very large single-participant functional magnetic resonance data that leverages the recent success of image-to-image transformation networks. This is achieved by exploiting spatial information obtained from retinotopic mappings across the visual system. More specifically, we first determine what position each voxel in a particular region of interest would represent in the visual field based on its corresponding receptive field location. Then, the 2D image representation of the brain activity on the visual field is passed to a fully convolutional image-to-image network trained to recover the original stimuli using VGG feature loss with an adversarial regularizer. In our experiments, we show that our method offers a significant improvement over existing video reconstruction techniques.

H-YOLO: A Single-Shot Ship Detection Approach Based on Region of Interest Preselected Network

Ship detection from high-resolution optical satellite images is still an important task that deserves optimal solutions. This paper introduces a novel high-resolution image network-based approach based on the preselection of a region of interest (RoI). This pre-selected network first identifies and extracts a region of interest from input images. In order to efficiently match ship candidates, the principle of our approach is to distinguish suspected areas from the images based on hue, saturation, value (HSV) differences between ships and the background. The whole approach is the basis of an experiment with a large ship dataset, consisting of Google Earth images and HRSC2016 datasets. The experiment shows that the H-YOLO network, which uses the same weight training from a set of remote sensing images, has a 19.01% higher recognition rate and a 16.19% higher accuracy than applying the you only look once (YOLO) network alone. After image preprocessing, the value of the intersection over union (IoU) is also greatly improved.

FISHEYE PHOTOGRAMMETRY TO GENERATE LOW-COST DTMS

In the archaeological practice, Digital Terrain Models (DTMs) and Digital Surface Models (DSMs) may be used to represent spatial information about the site by conveying information such as differences in levels, morphology of the terrain and movements of volumes during the excavation. Nowadays DTMs and DSMs can be easily obtained by image-based matching using low altitude aerial dataset acquired from a digital camera by means of a lifting device. In recent years, the spread of commercial multi-rotor unmanned aerial vehicles and their decreasing cost made low-altitude aerial photography even easier than before, where balloons, kites and telescopic masts would have been used instead. However, the use of drones is often forbidden by law, especially in the archaeological areas, and therefore a more traditional approach must to be adopted instead.This paper presents two different approaches adopted on the field to acquire the DTM of an archaeological excavation: the use of a pole held by a chest harness to lift a camera up to 3.5 m height fitted with a 20 mm wide angle lens; and a second solution that exploits ground-based fisheye photogrammetry. In general, an image network acquired from ground level is challenging due to: i) the poor coverage that can be obtained on the ground, ii) the large number of images that are required to cover large areas and consequently iii) the longer elaboration time that is required to process the data. The fisheye approach, however, proved to be more effective thanks to the more robust image network resulting both from the wider field of view and from the possibility to handle large datasets by downsampling the images and still retrieving strong key points. The main difference with the first system is that the monotonous images acquired by the 20 mm lens, very plain in texture, require working at full resolution in order to distinguish valid features in the sand.The final product of the tests carried out along this line in 2019 at Saqqara (Egypt) is a comprehensive DSM of the entire archaeological site with an accuracy of ~3 cm.

Reports of my demise are greatly exaggerated: $N$-subjettiness taggers take on jet images

We compare the performance of a convolutional neural network (CNN) trained on jet images with dense neural networks (DNNs) trained on nn-subjettiness variables to study the distinguishing power of these two separate techniques applied to top quark decays. We find that they perform almost identically and are highly correlated once jet mass information is included, which suggests they are accessing the same underlying information which can be intuitively understood as being contained in 4-, 5-, 6-, and 8-body kinematic phase spaces depending on the sample. This suggests both of these methods are highly useful for heavy object tagging and provides a tentative answer to the question of what the image network is actually learning.