In this paper, a novel approach to video temporal decomposition into semantic units, termed scenes, is presented. In contrast to previous temporal segmentation approaches that employ mostly low-level visual or audiovisual features, we introduce a technique that jointly exploits low-level and high-level features automatically extracted from the visual and the auditory channel. This technique is built upon the well-known method of the scene transition graph (STG), first by introducing a new STG approximation that features reduced computational cost, and then by extending the unimodal STG-based temporal segmentation technique to a method for multimodal scene segmentation. The latter exploits, among others, the results of a large number of TRECVID-type trained visual concept detectors and audio event detectors, and is based on a probabilistic merging process that combines multiple individual STGs while at the same time diminishing the need for selecting and fine-tuning several STG construction parameters. The proposed approach is evaluated on three test datasets, comprising TRECVID documentary films, movies, and news-related videos, respectively. The experimental results demonstrate the improved performance of the proposed approach in comparison to other unimodal and multimodal techniques of the relevant literature and highlight the contribution of high-level audiovisual features toward improved video segmentation to scenes.

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.

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.