60419 research outputs found
Sort by
Sketch Less for More: On-the-Fly Fine-Grained Sketch Based Image Retrieval
No full textFine-grained sketch-based image retrieval (FG-SBIR) addresses the problem of retrieving a particular photo instance given a user’s query sketch. Its widespread applicability is however hindered by the fact that drawing a sketch takes time, and most people struggle to draw a complete and faithful sketch. In this paper, we reformulate the conventional FG-SBIR framework to tackle these challenges, with the ultimate goal of retrieving the target photo with the least number of strokes possible. We further propose an on-the-fly design that starts retrieving as soon as the user starts drawing. To accomplish this, we devise a reinforcement learning based cross-modal retrieval framework that directly optimizes rank of the ground-truth photo over a complete sketch drawing episode. Additionally, we introduce a novel reward scheme that circumvents the problems related to irrelevant sketch strokes, and thus provides us with a more consistent rank list during the retrieval. We achievesuperiorearly-retrievalefficiencyoverstate-of-theartmethodsandalternativebaselinesontwopubliclyavailable fine-grained sketch retrieval datasets
Tradeoffs between gene expression, growth and phenotypic diversity in microbial populations
No full textBacterial cells have a limited number of resources that can be allocated for gene expression. The intracellular competition for these resources has an impact on the cell physiology. Bacteria have evolved mechanisms to optimize resource allocation in a variety of scenarios, showing a tradeoff between the resources used to maximise growth (e.g. ribosome synthesis) and the rest of cellular functions. Limitations in gene expression also play a role in generating phenotypic diversity, which is advantageous in fluctuating environments, at the expenses of decreasing growth rates. Our current understanding of these tradeoffs can be exploited for biotechnological applications benefiting from the selective manipulation of the allocation of resources
Indoor scene understanding from visual analysis of human activity
No full textVisual scene understanding studies the task of representing a captured scene in a manner emulating human-like understanding of that space. Considering indoor scenes are designed for human use and are utilised everyday, attaining this understanding is crucial for applications such as robotic mapping and navigation, smart home and security systems, and home healthcare and assisted living. However, although we as humans utilise such spaces in our day-to-day lives, analysis of human activity is not commonly applied towards enhancing indoor scene-level understanding. As such, the work presented in this thesis investigates the benefits of including human activity information in indoor scene understanding challenges, aiming to demonstrate its potential contributions, applications, and versatility.
The first contribution of this thesis utilises human activity to reveal scene regions occluded behind objects and clutter. Human poses recognised from a static sensor are projected into a top-down scene representation recording belief of human activity over time. This representation is applied to carve a volumetric scene map, initialised on captured depth, to expose the occupancy of hidden scene regions. An object detection approach exploits the revealed occluded scene occupancy to localise self-, partially-, and, significantly, fully-occluded objects. The second contribution extends the top-down activity representation to predict the functionality of major scene surfaces from human activity recognised in 360 degree video. A convolutional network is trained on simulated human activity to segment walkable, sittable, and interactable surfaces from the top-down perspective. This prediction is applied to construct a complete scene 3D approximation, with results showing scene structure and surface functionality are predicted well from human activity alone. Finally, this thesis investigates an association between the top-down functionality prediction and the captured visual scene. A new dataset capturing long-term human activity is introduced to train a model on combined activity and visual scene information. The model is trained to segment functional scene surfaces from the capture sensor perspective, with evaluation establishing that the introduction of human activity information can improve functional surface segmentation performance.
Overall, the work presented in this thesis demonstrates that analysis of human activity can be applied to enhance indoor scene understanding across various challenges, sensors, and representations. Assorted datasets are introduced alongside the major contributions to motivate further investigation into its application
Dynamic ball indentation for powder flow characterization
No full textIn industrial processing and manufacturing, characterizing the flowability of particulate solids is of particular importance both for reliable powder flow and for a consistent production rate. Shear testing is the most widely used method for powders subjected to moderate or high stresses, and under quasi-static conditions. However, this method is not suitable for measuring the powder flow properties occurring in dynamic systems, such as powder mixers and screw conveyors. In this study, the rheological behaviour of powders at high shear rates has been evaluated by the ball indentation method. The technique, which simply consists of dropping a ball onto a cylindrical bed of previously consolidated powder, directly measures the material hardness, which is related to the unconfined yield stress by the constraint factor. The impact of the ball on the bed is recorded with a high-speed camera to determine velocity and penetration depth. The hardness against the strain rate is considered for four different materials. Because of their difference in particle size, and by using a range of drop heights and a range of indenter densities, the intermediate regime of flow has been fully analyzed. Although hardness is constant in the quasi-static condition, it results to be strain rate dependent in the intermediate regime of flow. Finally, a predictive correlation that allows the operator to choose the best operating conditions for achieving the desired flow regime is proposed, and the unconfined yield strength of the materials is inferred
Mycobacterium bovis uses the ESX-1 Type VII secretion system to escape predation by the soil-dwelling amoeba Dictyostelium discoideum
No full textMycobacterium bovis is the causative agent of bovine tuberculosis and the predominant cause of zoonotic tuberculosis in people. Bovine tuberculosis occurs in farmed cattle but also in a variety of wild animals, which form a reservoir of infection. Although direct transmission of tuberculosis occurs between mammals, the low frequency of contact between different host species and abundant shedding of bacilli by infected animals suggests an infectious route via environmental contamination. Other intracellular pathogens that transmit via the environment deploy strategies to survive or exploit predation by environmental amoebae. To explore if M. bovis has this capability, we investigated its interactions with the soil and dung-dwelling amoeba, Dictyostelium discoideum. We demonstrated that M. bovis evades phagocytosis and destruction by D. discoideum and actively transits through the amoeba using the ESX-1 Type VII Secretion System as part of a programme of mechanisms, many of which have been co-opted as virulence factors in the mammalian host. This capacity of M. bovis to utilise an environmental stage between mammalian hosts may enhance its transmissibility. In addition, our data provide molecular evidence to support an evolutionary role for amoebae as training grounds for the pathogenic M. tuberculosis complex
A Novel Unipolar Transmission Scheme for Visible Light Communication
No full textThis paper proposes a novel unipolar transceiver for visible light communication (VLC) by using
orthogonal waveforms. The main advantage of our proposed scheme over most of the existing unipolar
schemes in the literature is that the polarity of the real-valued orthogonal frequency division multiplexing
(OFDM) sample determines the pulse shape of the continuous-time signal and thus, the unipolar
conversion is performed directly in the analog instead of the digital domain. Therefore, our proposed
scheme does not require any direct current (DC) biasing or clipping as it is the case with existing schemes
in the literature. The bit error rate (BER) performance of our proposed scheme is analytically derived
and its accuracy is verified by using Matlab simulations. Simulation results also substantiate the potential
performance gains of our proposed scheme against the state-of-the-art OFDM-based systems in VLC; it
indicates that the absence of DC shift and clipping in our scheme supports more reliable communication
and outperforms the asymmetrically clipped optical-OFDM (ACO-OFDM), DC optical-OFDM (DCOOFDM)
and unipolar-OFDM (U-OFDM) schemes. For instance, our scheme outperforms ACO-OFDM
by at least 3 dB (in terms of signal to noise ratio) at a target BER of 1
Nutrition, body composition, inflammation and haemoglobin status among haemodialysis patients on Erythropoietin maintenance therapy
No full textEnd stage renal disease (ESRD) is associated with several physiological and metabolic changes which together cause poor health and impair patients’ quality of life. Anaemia is one of the most common metabolic consequences of renal failure and is highly prevalent among ESRD patients. Renal anaemia has a significant negative impact not only on patients’ health outcomes but also poses a heavy economic burden on the health system due to the high cost of treatment. Interactions between nutrition, anaemia status or/and response to anaemia therapies have been suggested. This programme of work aimed to investigate anaemia and nutritional status (including body composition) among haemodialysis (HD) patients and to study how nutrition status and body composition modify patients’ anaemia status and their response to the hormone erythropoietin (EPO). The first two sections of this project were cross-sectional studies conducted to characterise the health status of patients. Firstly, health outcomes were compared between the UK (n = 101) and Tanzanian (n = 96) HD patients after which a more in depth study was conducted in Tanzania only (n = 77). These are presented in chapter 2 and 3 respectively. Both studies showed that Tanzanian HD patients had increased risk of low haemoglobin (Hb) levels, inadequate nutrition and inflammation. The findings showed that body fat and muscle mass were significantly and positively associated with Hb and EPO response. These investigations were then extended to the UK via recruitment and assessment of a cohort of UK adults HD patients on EPO treatment (n = 41). The UK study was conducted longitudinally and included additional markers of anaemia and nutritional status to further assess the relationship previously observed and in an attempt to uncover the mechanisms involved. The findings showed that longitudinally, body fat, significantly and positively influenced Hb concentration independent of EPO type and gender. Equally, there was a positive association between leptin hormone and Hb concentrations, suggesting a potential link and possible mechanism for the observed association between body fat and Hb levels. Finally, building on this work, an exercise intervention of low intensity (walking + intradialytic resistance training) was adopted and conducted as a randomized controlled pilot and feasibility study. The aims were to investigate if such an intervention was feasible and acceptable to patients and staff and capable of inducing favourable changes in patients’ body composition (body fat and muscle), and if so, how any such changes would modify patients response to EPO and their Hb status. The eight week study was unable to achieve significant changes in body composition or iron markers. However, there was a slight improvement in exercise performance and lower body strength assessed through the sit to stand test. To our knowledge, the low intensity exercise protocol used, combining walking + intradialytic resistance training, has not been previously tested in the dialysis setting and offers an acceptable intervention for further research.
This programme of work has demonstrated that good nutrition helps improve the EPO response and Hb levels of HD patients. A flexible and low intensity exercise programme involving walking and intradialytic resistance training is feasible and practical in the presence of a physiotherapist. However eight weeks of low intensity exercise is not sufficient to induce changes in body composition or EPO response in HD patients but may help improve their physical functioning and so warrants further investigation as a more cost effective alternative to cycle-based interventions
State-Dependent Service Rates in Make-to-Order Shops: An Assessment by Simulation
No full textMost literature on make-to-order shops assumes that service rates are independent of the system state. In practice however, the service rate is often dependent on the workload level experienced by the worker. While a body of knowledge on state-dependent service rates exists, the available literature has not given sufficient attention to make-to-order shops, which are often characterized by complex routings and defined due dates, which means delivery performance becomes a major concern. This study uses simulation to assess the performance impact of state-dependent service rates under different degrees of routing directedness. We show that including information on the load upstream of a station when making service rate adjustments has the potential to improve performance compared to considering the load directly queuing at a station only, as has been the case in previous research on state-dependent service rates. Moreover, using the same threshold to trigger service rate adjustments at each station in shops with directed routings leads to higher service rates at upstream stations. This service rate imbalance can be avoided by using different triggering thresholds for upstream and downstream stations. Further, and most importantly, we show that although speeding up behavior during high load periods significantly improves performance, if worker fatigue leads to a decrease in the service rate in response to the initial increase then performance may in fact deteriorate
Data association and uncertainty pruning for tracks determined on short arcs
No full textWhen building a space catalogue, it is necessary to acquire multiple observations of the same object for the estimated state to be considered meaningful. A first concern is then to establish whether different sets of observations belong to the same object, which is the association problem. Due to illumination constraints and adopted observation strategies, small objects may be detected on short arcs, which contain little information about the curvature of the orbit. Thus, a single detection is usually of little value in determining the orbital state due to the very large associated uncertainty. In this work, we propose a method that both recognizes associated observations and sequentially reduces the solution uncertainty when two or more sets of observations are associated. The six-dimensional (6D) association problem is addressed as a cascade of 2D and 4D optimization problems. The performance of the algorithm is assessed using objects in geostationary Earth orbit, with observations spread over short arcs
Decay properties of 22Ne + � resonances and their impact on s-process nucleosynthesis
No full textThe astrophysical s-process is one of the two main processes forming elements heavier
than iron. A key outstanding uncertainty surrounding s-process nucleosynthesis
is the neutron flux generated by the 22Ne(�; n)25Mg reaction during the He-core and
C-shell burning phases of massive stars. This reaction, as well as the competing
22Ne(�;)26Mg reaction, is not well constrained in the important temperature regime
from �0:2–0:4 GK, owing to uncertainties in the nuclear properties of resonances lying
within the Gamow window. To address these uncertainties, we have performed
a new measurement of the 22Ne(6Li; d)26Mg reaction in inverse kinematics, detecting
the outgoing deuterons and 25;26Mg recoils in coincidence. We have established a
new n=
decay branching ratio of 1:14(26) for the key Ex = 11:32 MeV resonance in
26Mg, which results in a new (�; n) strength for this resonance of 42(11) �eV when
combined with the well-established (�;
) strength of this resonance. We have also determined
new upper limits on the � partial widths of neutron-unbound resonances at
Ex = 11:112; 11:163, 11:169, and 11:171 MeV. Monte-Carlo calculations of the stellar 22Ne(�; n)25Mg and 22Ne(�;
)26Mg rates, which incorporate these results, indicate
that both rates are substantially lower than previously thought in the temperature range
from �0:2–0:4 GK
Data sources:
Access Repository Dashboard
Do you manage Surrey Research Insight? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!