No abstract

No abstract

This work introduces a method that combines remote sensing and deep learning into a framework that is tailored for accurate, reliable and efficient counting and sizing of plants in aerial images. The investigated task focuses on two low-density crops, potato and lettuce. This double objective of counting and sizing is achieved through the detection and segmentation of individual plants by fine-tuning an existing deep learning architecture called Mask R-CNN. This paper includes a thorough discussion on the optimal parametrisation to adapt the Mask R-CNN architecture to this novel task. As we examine the correlation of the Mask R-CNN performance to the annotation volume and granularity (coarse or refined) of remotely sensed images of plants, we conclude that transfer learning can be effectively used to reduce the required amount of labelled data. Indeed, a previously trained Mask R-CNN on a low-density crop can improve performances after training on new crops. Once trained for a given crop, the Mask R-CNN solution is shown to outperform a manually-tuned computer vision algorithm. Model performances are assessed using intuitive metrics such as Mean Average Precision (mAP) from Intersection over Union (IoU) of the masks for individual plant segmentation and Multiple Object Tracking Accuracy (MOTA) for detection. The presented model reaches an mAP of 0.418 for potato plants and 0.660 for lettuces for the individual plant segmentation task. In detection, we obtain a MOTA of 0.781 for potato plants and 0.918 for lettuces.

No abstract

Cloud masking is of central importance to the Earth Observation community. This paper deals with the problem of detecting clouds in visible and multispectral imagery from high-resolution satellite cameras. Recently, Machine Learning has offered promising solutions to the problem of cloud masking, allowing for more flexibility than traditional thresholding techniques, which are restricted to instruments with the requisite spectral bands. However, few studies use multi-scale features (as in, a combination of pixel-level and spatial) whilst also offering compelling experimental evidence for real-world performance. Therefore, we introduce CloudFCN, based on a Fully Convolutional Network architecture, known as U-net, which has become a standard Deep Learning approach to image segmentation. It fuses the shallowest and deepest layers of the network, thus routing low-level visible content to its deepest layers. We offer an extensive range of experiments on this, including data from two high-resolution sensors—Carbonite-2 and Landsat 8—and several complementary tests. Owing to a variety of performance-enhancing design choices and training techniques, it exhibits state-of-the-art performance where comparable to other methods, high speed, and robustness to many different terrains and sensor types.

The first high-resolution Digital Terrain Model (DTM) of the entire South Pole of Mars has been produced. A modified version (Kim and Muller, 2009) of a NASA-VICAR-based pipeline developed by DLR (German Aerospace Centre) and JPL (Jet Propulsion Laboratory) has been employed with image matching based on the Gotcha (Gruen-Otto-Chau) algorithm (Shin and Muller, 2012) with a specialised setup for the polar region. DTM products have been produced with more than twice the resolution (50m/pixel) of the gridded Mars Orbiter Laser Altimeter (MOLA) 512 pixels/degree (112 m/pixel) over the South Polar Residual Cap (SPRC) and the Mars South Polar region (82 o -90 o S) in MOLA and areoid reference. The accuracy of the HRSC orbital DTMs are compared against a MOLA reference with good results. HRSC orthorectified strip images from 12.5-50m have also been produced from the base DTMs and these have been processed into a 12.5m mosaic. HRSC strip products are currently being assessed as base images for automatic co-registration of thousands of high-resolution images, making them geometrically consistent with the surface conditions imaged by HRSC. In some cases, CTX DTMs have been automatically produced and co-registered to the HRSC image strips and these, in turn, are being employed for automated co-registration of higher-resolution images.

In this article, we discuss the potential benefits, the requirements and the challenges involved in patent image retrieval and subsequently, we propose a framework that encompasses advanced image analysis and indexing techniques to address the need for content-based patent image search and retrieval. The proposed framework involves the application of document image pre-processing, image feature and textual metadata extraction in order to support effectively content-based image retrieval in the patent domain. To evaluate the capabilities of our proposal, we implemented a patent image search engine. Results based on a series of interaction modes, comparison with existing systems and a quantitative evaluation of our engine provide evidence that image processing and indexing technologies are currently mature to be integrated in real-world patent retrieval applications.

Thirty consecutive patients who had suffered unstable fractures and dislocations of the thoracolumbar spine mostly associated with neurologic impairment and bony encroachment on the spinal canal were treated either with Harrington distraction rods combined with sublaminar wires or with the Zielke-VDS device. These patients were subsequently assessed for neurologic outcome, spinal canal clearance, sagittal and coronal spinal deformity correction preoperatively and postoperatively with a minimum follow-up of 26 months. In the follow-up evaluation, the patients who underwent surgery with Harrington rods showed an overall improvement of their neurologic function of 90.9%, whereas all patients who underwent the Zielke operation improved. Preoperatively, positive correlations were found between the level of injury and Frankel grades; the cord lesion tended to demonstrate more severe neurologic deficit when compared with cauda equina ones (P < 0.001). Furthermore, dislocation accompanying the injury resulted in a more severe neurological deficit (P < 0.05). Harrington rods and Zielke device offer sufficient initial correction of the frontal spinal deformity but did not significantly either restore or maintain sagittal plane alignment. The Harrington series showed an overall improvement of the segmental kyphosis of 26% (NS), with a subsequent loss of correction of 7.38% (NS) on the follow-up observation. The Zielke device produced an immediate, much better correction of the segmental posttraumatic kyphosis of 45% (NS), but a loss of correction of 22.9% (NS) was measured in the follow-up evaluation. Correction of the anterior and posterior vertebral height was shown to be better for the Zielke patient group.(ABSTRACT TRUNCATED AT 250 WORDS)