Soil organic matter molecular composition with long-term detrital alterations is controlled by site-specific forest properties.

Forest ecosystems are important global soil carbon (C) reservoirs, but their capacity to sequester C is susceptible to climate change factors that alter the quantity and quality of C inputs. To better understand forest soil C responses to altered C inputs, we integrated three molecular composition published data sets of soil organic matter (SOM) and soil microbial communities for mineral soils after 20 years of detrital input and removal treatments in two deciduous forests: Bousson Forest (BF), Harvard Forest (HF), and a coniferous forest: H.J. Andrews Forest (HJA). Soil C turnover times were estimated from radiocarbon measurements and compared with the molecular-level data (based on nuclear magnetic resonance and specific analysis of plant- and microbial-derived compounds) to better understand how ecosystem properties control soil C biogeochemistry and dynamics. Doubled aboveground litter additions did not increase soil C for any of the forests studied likely due to long-term soil priming. The degree of SOM decomposition was higher for bacteria-dominated sites with higher nitrogen (N) availability while lower for the N-poor coniferous forest. Litter exclusions significantly decreased soil C, increased SOM decomposition state, and led to the adaptation of the microbial communities to changes in available substrates. Finally, although aboveground litter determined soil C dynamics and its molecular composition in the coniferous forest (HJA), belowground litter appeared to be more influential in broadleaf deciduous forests (BH and HF). This synthesis demonstrates that inherent ecosystem properties regulate how soil C dynamics change with litter manipulations at the molecular-level. Across the forests studied, 20 years of litter additions did not enhance soil C content, whereas litter reductions negatively impacted soil C concentrations. These results indicate that soil C biogeochemistry at these temperate forests is highly sensitive to changes in litter deposition, which are a product of environmental change drivers.

Weakly Supervised Reinforcement Learning for Autonomous Highway Driving via Virtual Safety Cages.

The use of neural networks and reinforcement learning has become increasingly popular in autonomous vehicle control. However, the opaqueness of the resulting control policies presents a significant barrier to deploying neural network-based control in autonomous vehicles. In this paper, we present a reinforcement learning based approach to autonomous vehicle longitudinal control, where the rule-based safety cages provide enhanced safety for the vehicle as well as weak supervision to the reinforcement learning agent. By guiding the agent to meaningful states and actions, this weak supervision improves the convergence during training and enhances the safety of the final trained policy. This rule-based supervisory controller has the further advantage of being fully interpretable, thereby enabling traditional validation and verification approaches to ensure the safety of the vehicle. We compare models with and without safety cages, as well as models with optimal and constrained model parameters, and show that the weak supervision consistently improves the safety of exploration, speed of convergence, and model performance. Additionally, we show that when the model parameters are constrained or sub-optimal, the safety cages can enable a model to learn a safe driving policy even when the model could not be trained to drive through reinforcement learning alone.

Chlorination of soil-derived dissolved organic matter: Long term nitrogen deposition does not increase terrestrial precursors of toxic disinfection byproducts.

Terrestrial dissolved organic matter (DOM) in forested watersheds is a known precursor of disinfection byproducts (DBPs) in drinking water. Although the characteristics of terrestrial DOM may change with increasing nitrogen (N) deposition in forests, how these changes alter formation potential and toxicity of DBPs remains unexplored. We analyzed the speciation and toxicity of DBPs from chlorination of DOM derived from soils (O, A, and B horizons) in an experimental temperate forest with 22 years of N addition. With long-term N addition, the DOM reactivity toward the formation of trihalomethanes (from 27.7-51.8 to 22.8-31.1 µg/mg-dissolved organic carbon (DOC)) and chloral hydrate (from 1.25-1.63 to 1.14-1.36 µg/mg-DOC) decreased, but that toward the formation of haloketones increased (from 0.23-0.26 to 0.26-0.33 µg/mg-DOC). The DOM reactivity toward the formation of haloacetonitriles was increased in the deeper soil but reduced in the surface soil. The DBP formation potential of DOM draining from a certain area of forest soils (in µg-DBP/m2-soil) was estimated to be reduced by 20.3% for trihalomethanes and increased by 37.5% for haloketones and have minor changes for haloacetonitriles and chloral hydrate (both <7%). Furthermore, the DBPs from chlorination of the soil-derived DOM showed lowered microtoxicity with N addition possibly due to reduced brominated DBP formation. Overall, this study highlights that N deposition may not increase drinking water toxicity through altering terrestrial DOM characteristics.

Weakly Supervised Learning with Multi-Stream CNN-LSTM-HMMs to Discover Sequential Parallelism in Sign Language Videos.

In this work we present a new approach to the field of weakly supervised learning in the video domain. Our method is relevant to sequence learning problems which can be split up into sub-problems that occur in parallel. Here, we experiment with sign language data. The approach exploits sequence constraints within each independent stream and combines them by explicitly imposing synchronisation points to make use of parallelism that all sub-problems share. We do this with multi-stream HMMs while adding intermediate synchronisation constraints among the streams. We embed powerful CNN-LSTM models in each HMM stream following the hybrid approach. This allows the discovery of attributes which on their own lack sufficient discriminative power to be identified. We apply the approach to the domain of sign language recognition exploiting the sequential parallelism to learn sign language, mouth shape and hand shape classifiers. We evaluate the classifiers on three publicly available benchmark data sets featuring challenging real-life sign language with over 1,000 classes, full sentence based lip-reading and articulated hand shape recognition on a fine-grained hand shape taxonomy featuring over 60 different hand shapes. We clearly outperform the state-of-the-art on all data sets and observe significantly faster convergence using the parallel alignment approach.

HARD-PnP: PnP Optimization Using a Hybrid Approximate Representation.

This paper proposes a Hybrid Approximate Representation (HAR) based on unifying several efficient approximations of the generalized reprojection error (which is known as the gold standard for multiview geometry). The HAR is an over-parameterization scheme where the approximation is applied simultaneously in multiple parameter spaces. A joint minimization scheme "HAR-Descent" can then solve the PnP problem efficiently, while remaining robust to approximation errors and local minima. The technique is evaluated extensively, including numerous synthetic benchmark protocols and the real-world data evaluations used in previous works. The proposed technique was found to have runtime complexity comparable to the fastest O(n) techniques, and up to 10 times faster than current state of the art minimization approaches. In addition, the accuracy exceeds that of all 9 previous techniques tested, providing definitive state of the art performance on the benchmarks, across all 90 of the experiments in the paper and supplementary material, which can be found on the Computer Society Digital Library at http://doi.ieeecomputersociety.org/10.1109/TPAMI.2018.2806446.

The detrital input and removal treatment (DIRT) network: Insights into soil carbon stabilization.

Ecological research networks functioning across climatic and edaphic gradients are critical for improving predictive understanding of biogeochemical cycles at local through global scales. One international network, the Detrital Input and Removal Treatment (DIRT) Project, was established to assess how rates and sources of plant litter inputs influence accumulations or losses of organic matter in forest soils. DIRT employs chronic additions and exclusions of aboveground litter inputs and exclusion of root ingrowth to permanent plots at eight forested and two shrub/grass sites to investigate how soil organic matter (SOM) dynamics are influenced by plant detrital inputs across ecosystem and soil types. Across the DIRT network described here, SOM pools responded only slightly, or not at all, to chronic doubling of aboveground litter inputs. Explanations for the slow or even negative response of SOM to litter additions include increased decomposition of new inputs and priming of old SOM. Evidence of priming includes increased soil respiration in litter addition plots, decreased dissolved organic carbon (DOC) output from increased microbial activity, and biochemical markers in soil indicating enhanced SOM degradation. SOM pools decreased in response to chronic exclusion of aboveground litter, which had a greater effect on soil C than did excluding roots, providing evidence that root-derived C is not more critical than aboveground litter C to soil C sequestration. Partitioning of belowground contributions to total soil respiration were predictable based on site-level soil C and N as estimates of site fertility; contributions to soil respiration from root respiration were negatively related to soil fertility and inversely, contributions from decomposing aboveground litter in soil were positively related to site fertility. The commonality of approaches and manipulations across the DIRT network has provided greater insights into soil C cycling than could have been revealed at a single site.

Long-term litter manipulation alters soil organic matter turnover in a temperate deciduous forest.

Understanding soil organic matter (OM) biogeochemistry at the molecular-level is essential for assessing potential impacts from management practices and climate change on shifts in soil carbon storage. Biomarker analyses and nuclear magnetic resonance (NMR) spectroscopy were used in an ongoing detrital input and removal treatment experiment in a temperate deciduous forest in Pennsylvania, USA, to examine how above- and below-ground plant inputs control soil OM quantity and quality at the molecular-level. From plant material to surface soils, the free acyclic lipids and cutin, suberin, and lignin biomarkers were preferentially retained over free sugars and free cyclic lipids. After 20years of above-ground litter addition (Double Litter) or exclusion (No Litter) treatments, soil OM composition was relatively more degraded, as revealed by solid-state 13C NMR spectroscopy. Under Doubled Litter inputs, soil carbon and phospholipid fatty acid (PLFA) concentrations were unchanged, suggesting that the current OM degradation status is a reflection of microbial-mediated degradation that occurred prior to the 20-year sampling campaign. Soil OM degradation was higher in the No Litter treatments, likely due to the decline in fresh, above-ground litter inputs over time. Furthermore, root and root and litter exclusion treatments (No Roots and No Inputs, respectively) both significantly reduced free sugars and PLFAs and increased preservation of suberin-derived compounds. PLFA stress ratios and the low N-acetyl resonances from diffusion edited 1H NMR also indicate substrate limitations and reduced microbial biomass with these treatments. Overall, we highlight that storage of soil carbon and its biochemical composition do not linearly increase with plant inputs because the microbial processing of soil OM is also likely altered in the studied forest.

TMAGIC: A Model-Free 3D Tracker.

Significant effort has been devoted within the visual tracking community to rapid learning of object properties on the fly. However, state-of-the-art approaches still often fail in cases such as rapid out-of-plane rotation, when the appearance changes suddenly. One of the major contributions of this paper is a radical rethinking of the traditional wisdom of modeling 3D motion as appearance changes during tracking. Instead, 3D motion is modeled as 3D motion. This intuitive but previously unexplored approach provides new possibilities in visual tracking research. First, 3D tracking is more general, as large out-of-plane motion is often fatal for 2D trackers, but helps 3D trackers to build better models. Second, the tracker's internal model of the object can be used in many different applications and it could even become the main motivation, with tracking supporting reconstruction rather than vice versa. This effectively bridges the gap between visual tracking and structure from motion. A new benchmark data set of sequences with extreme out-of-plane rotation is presented and an online leader-board offered to stimulate new research in the relatively underdeveloped area of 3D tracking. The proposed method, provided as a baseline, is capable of successfully tracking these sequences, all of which pose a considerable challenge to 2D trackers (error reduced by 46%).

Hollywood 3D: What are the Best 3D Features for Action Recognition?

Action recognition "in the wild" is extremely challenging, particularly when complex 3D actions are projected down to the image plane, losing a great deal of information. The recent growth of 3D data in broadcast content and commercial depth sensors, makes it possible to overcome this. However, there is little work examining the best way to exploit this new modality. In this paper we introduce the Hollywood 3D benchmark, which is the first dataset containing "in the wild" action footage including 3D data. This dataset consists of 650 stereo video clips across 14 action classes, taken from Hollywood movies. We provide stereo calibrations and depth reconstructions for each clip. We also provide an action recognition pipeline, and propose a number of specialised depth-aware techniques including five interest point detectors and three feature descriptors. Extensive tests allow evaluation of different appearance and depth encoding schemes. Our novel techniques exploiting this depth allow us to reach performance levels more than triple those of the best baseline algorithm using only appearance information. The benchmark data, code and calibrations are all made available to the community.

Texture-Independent Long-Term Tracking Using Virtual Corners.

Long-term tracking of an object, given only a single instance in an initial frame, remains an open problem. We propose a visual tracking algorithm, robust to many of the difficulties that often occur in real-world scenes. Correspondences of edge-based features are used, to overcome the reliance on the texture of the tracked object and improve invariance to lighting. Furthermore, we address long-term stability, enabling the tracker to recover from drift and to provide redetection following object disappearance or occlusion. The two-module principle is similar to the successful state-of-the-art long-term TLD tracker; however, our approach offers better performance in benchmarks and extends to cases of low-textured objects. This becomes obvious in cases of plain objects with no texture at all, where the edge-based approach proves the most beneficial. We perform several different experiments to validate the proposed method. First, results on short-term sequences show the performance of tracking challenging (low textured and/or transparent) objects that represent failure cases for competing the state-of-the-art approaches. Second, long sequences are tracked, including one of almost 30 000 frames, which, to the best of our knowledge, is the longest tracking sequence reported to date. This tests the redetection and drift resistance properties of the tracker. Finally, we report the results of the proposed tracker on the VOT Challenge 2013 and 2014 data sets as well as on the VTB1.0 benchmark, and we show relative performance of the tracker compared with its competitors. All the results are comparable with the state of the art on sequences with textured objects and superior on non-textured objects. The new annotated sequences are made publicly available.

Scene particles: unregularized particle-based scene flow estimation.

In this paper, an algorithm is presented for estimating scene flow, which is a richer, 3D analog of optical flow. The approach operates orders of magnitude faster than alternative techniques and is well suited to further performance gains through parallelized implementation. The algorithm employs multiple hypotheses to deal with motion ambiguities, rather than the traditional smoothness constraints, removing oversmoothing errors and providing significant performance improvements on benchmark data, over the previous state of the art. The approach is flexible and capable of operating with any combination of appearance and/or depth sensors, in any setup, simultaneously estimating the structure and motion if necessary. Additionally, the algorithm propagates information over time to resolve ambiguities, rather than performing an isolated estimation at each frame, as in contemporary approaches. Approaches to smoothing the motion field without sacrificing the benefits of multiple hypotheses are explored, and a probabilistic approach to occlusion estimation is demonstrated, leading to 10 and 15 percent improved performance, respectively. Finally, a data-driven tracking approach is described, and used to estimate the 3D trajectories of hands during sign language, without the need to model complex appearance variations at each viewpoint.

Root stress and nitrogen deposition: consequences and research priorities.

Stress within tree roots may influence whole-tree responses to nutrient deficiencies or toxic ion accumulation, but the mechanisms that govern root responses to the belowground chemical environment are poorly quantified. Currently, root production is modeled using rates of forest production and stoichiometry, but this approach alone may be insufficient to forecast variability in forest responses when physical and chemical stressors alter root lifespan, rooting depth or mycorrhizal colonization directly. Here, we review key research priorities for improving predictions of tree responses to changes in the belowground biogeochemical environment resulting from nitrogen deposition, including: limits of the optimum allocation paradigm, root physiological stress and lifespan, contingency effects that determine threshold responses across broad gradients, coupled water-biogeochemical interactions on roots, mycorrhizal dynamics that mediate root resilience and model frameworks to better simulate root feedbacks to aboveground function. We conclude that models incorporating physiological feedbacks, dynamic responses to coupled stressors, mycorrhizal interactions, and which challenge widely-accepted notions of optimum allocation, can elucidate potential thresholds of tree responses to biogeochemical stressors. Emphasis on comparative studies across species and environmental gradients, and which incorporates insights at the cellular and ecosystem level, is critical for forecasting whole-tree responses to altered biogeochemical landscapes.

Action recognition using mined hierarchical compound features.

The field of Action Recognition has seen a large increase in activity in recent years. Much of the progress has been through incorporating ideas from single-frame object recognition and adapting them for temporal-based action recognition. Inspired by the success of interest points in the 2D spatial domain, their 3D (space-time) counterparts typically form the basic components used to describe actions, and in action recognition the features used are often engineered to fire sparsely. This is to ensure that the problem is tractable; however, this can sacrifice recognition accuracy as it cannot be assumed that the optimum features in terms of class discrimination are obtained from this approach. In contrast, we propose to initially use an overcomplete set of simple 2D corners in both space and time. These are grouped spatially and temporally using a hierarchical process, with an increasing search area. At each stage of the hierarchy, the most distinctive and descriptive features are learned efficiently through data mining. This allows large amounts of data to be searched for frequently reoccurring patterns of features. At each level of the hierarchy, the mined compound features become more complex, discriminative, and sparse. This results in fast, accurate recognition with real-time performance on high-resolution video. As the compound features are constructed and selected based upon their ability to discriminate, their speed and accuracy increase at each level of the hierarchy. The approach is tested on four state-of-the-art data sets, the popular KTH data set to provide a comparison with other state-of-the-art approaches, the Multi-KTH data set to illustrate performance at simultaneous multiaction classification, despite no explicit localization information provided during training. Finally, the recent Hollywood and Hollywood2 data sets provide challenging complex actions taken from commercial movie sequences. For all four data sets, the proposed hierarchical approach outperforms all other methods reported thus far in the literature and can achieve real-time operation.

Robust Facial Feature Tracking Using Shape-Constrained Multiresolution-Selected Linear Predictors.

This paper proposes a learned data-driven approach for accurate, real-time tracking of facial features using only intensity information. The task of automatic facial feature tracking is nontrivial since the face is a highly deformable object with large textural variations and motion in certain regions. Existing works attempt to address these problems by either limiting themselves to tracking feature points with strong and unique visual cues (e.g., mouth and eye corners) or by incorporating a priori information that needs to be manually designed (e.g., selecting points for a shape model). The framework proposed here largely avoids the need for such restrictions by automatically identifying the optimal visual support required for tracking a single facial feature point. This automatic identification of the visual context required for tracking allows the proposed method to potentially track any point on the face. Tracking is achieved via linear predictors which provide a fast and effective method for mapping pixel intensities into tracked feature position displacements. Building upon the simplicity and strengths of linear predictors, a more robust biased linear predictor is introduced. Multiple linear predictors are then grouped into a rigid flock to further increase robustness. To improve tracking accuracy, a novel probabilistic selection method is used to identify relevant visual areas for tracking a feature point. These selected flocks are then combined into a hierarchical multiresolution LP model. Finally, we also exploit a simple shape constraint for correcting the occasional tracking failure of a minority of feature points. Experimental results show that this method performs more robustly and accurately than AAMs, with minimal training examples on example sequences that range from SD quality to Youtube quality. Additionally, an analysis of the visual support consistency across different subjects is also provided.

Decentralized sensor fusion for Ubiquitous Networking Robotics in Urban Areas.

In this article we explain the architecture for the environment and sensors that has been built for the European project URUS (Ubiquitous Networking Robotics in Urban Sites), a project whose objective is to develop an adaptable network robot architecture for cooperation between network robots and human beings and/or the environment in urban areas. The project goal is to deploy a team of robots in an urban area to give a set of services to a user community. This paper addresses the sensor architecture devised for URUS and the type of robots and sensors used, including environment sensors and sensors onboard the robots. Furthermore, we also explain how sensor fusion takes place to achieve urban outdoor execution of robotic services. Finally some results of the project related to the sensor network are highlighted.

Estimating the joint statistics of images using nonparametric windows with application to registration using mutual information.

Recently, the Nonparametric (NP) Windows has been proposed to estimate the statistics of real 1D and 2D signals. NP Windows is accurate, because it is equivalent to sampling images at a high (infinite) resolution for an assumed interpolation model. This paper extends the proposed approach to consider joint distributions of image-pairs. Second, Green's Theorem is used to simplify the previous NP Windows algorithm. Finally, a resolution-aware NP Windows algorithm is proposed to improve robustness to relative scaling between an image pair. Comparative testing of 2D image registration was performed using translation-only and affine transformations. Although more expensive than other methods, NP Windows frequently demonstrated superior performance for bias (distance between ground truth and global maximum) and frequency of convergence. Unlike other methods, the number of samples and the number of bins have little effect on NP Windows and the prior selection of a kernel is not required.

Mutual information for Lucas-Kanade Tracking (MILK): an inverse compositional formulation.

Mutual Information (MI) is popular for registration via function optimisation. This work proposes an inverse compositional formulation of MI for Levenberg-Marquardt optimisation. This yields a constant Hessian, which may be pre-computed. Speed improvements of 15% were obtained, with convergence accuracies similar to those of the standard formulation.

Tail profile: a more accurate system for analyzing DNA damage using the Comet assay.

The Comet assay (single cell gel electrophoresis assay) measures DNA strand breaks in individual cells. In the assay cells are embedded in agarose, lysed, and electrophoresed under low voltage, allowing migration of damaged DNA. The DNA is stained and subsequently viewed with an epifluorescent microscope. If DNA damage has occurred the electrophoresed DNA fragments appear as a diffuse tail behind the nucleus known as a "comet". Many computer-aided analysis systems are currently in use to quantify the amount of DNA damage that is represented by a comet image. Here, we present a novel method of analysis known as "tail profile". This method of analysis provides several advantages over currently employed methods, which rely primarily on the "tail moment" method of analysis. We compared the amount of DNA damage reported from both the tail profile and tail moment methods of analysis and observed a 26% (P<0.0001) increase in damage detected by tail profile across the 10-25 microm range of tail length, where the majority of the relevant comet data is concentrated. We further report that this increase in sensitivity is not only limited to assessing DNA damage, but also to gathering data from DNA repair assays. Furthermore, we demonstrate increased functionality and extended data analysis capabilities with the use of a compressed collection of images called a "comet chip" and through a visual representation of data called a "profile plot". Use of the custom macros enabled us to detect an unexpected characteristic of the electrophoretic profile, giving us novel insight into the nature of comet analysis. In addition to the increased analytical sensitivity proffered by this system, the tail profile macros are upgradeable and platform independent.