Food recognition using neural network classifier and multiple hypotheses image segmentation

ABSTRACT

This paper proposes the multiple-hypotheses image segmentation and feed-forward neural network classifier for food recognition to improve the performance. Initially, the food or meal image is given as input. Then, the segmentation is applied to identify the regions, where a particular food item is located using salient region detection, multi-scale segmentation, and fast rejection. Then, the features of every food item are extracted by the global feature and local feature extraction. After the features are obtained, the classification is performed for each segmented region using a feed-forward neural network model. Finally, the calorie value is computed with the aid of (i) food volume and (ii) calorie and nutrition measure based on mass value. The experimental results and performance evaluation are validated. The outcome of the proposed method attains 0.947 for Macro Average Accuracy (MAA) and 0.959 for Standard Accuracy (SA), which provides better classification performance.     

Tracking influent inorganic suspended solids through wastewater treatment plants.

From an experimental and theoretical investigation of the continuity of influent inorganic suspended solids (ISS) along the links connecting the primary settling tank (PST), fully aerobic or N removal activated sludge (AS) and anaerobic and aerobic sludge digestion unit operations, it was found that the influent wastewater (fixed) ISS concentration is conserved through primary sludge anaerobic digestion, activated sludge and aerobic digestion unit operations. However, the measured ISS flux at different stages through a series of wastewater treatment plant (WWTP) unit operations is not equal to the influent ISS flux, because the ordinary heterotrophic organisms (OHO) biomass contributes to the ISS flux by differing amounts depending on the active fraction of the VSS solids at that stage.     

Strain hardening due to deformation twinning in α-titanium: Mechanisms

Novel experiments were conducted to elucidate the effect of deformation twinning on the mechanical response of high-purity α-titanium deformed at room temperature. Orientation-imaging microscopy (OIM), microhardness, and nanohardness evaluations were employed in conjunction with optical microscopy and quasi-static compression testing to obtain insight into the deformation mechanisms. Hardness measurements revealed that the newly formed deformation twins were harder than the matrix. This observation is perhaps the first experimental evidence for the Basinski mechanism for hardening associated with twinning, arising from the transition of glissile dislocations to a sessile configuration upon the lattice reorientation by twinning shear. This work also provided direct evidence for two competing effects of deformation twinning on the overall stress-strain response: (1) hardening via both a reduction of the effective slip length (Hall-Petch effect) and an increase in the hardness of twinned regions (Basinski mechanism) and (2) softening due to the lattice reorientation of the twinned regions.     

Word length and lexical activation: Longer is better.

Many models of spoken word recognition posit the existence of lexical and sublexical representations, with excitatory and inhibitory mechanisms used to affect the activation levels of such representations. Bottom-up evidence provides excitatory input, and inhibition from phonetically similar representations leads to lexical competition. In such a system, long words should produce stronger lexical activation than short words, for 2 reasons: Long words provide more bottom-up evidence than short words, and short words are subject to greater inhibition due to the existence of more similar words. Four experiments provide evidence for this view. In addition, reaction-time-based partitioning of the data shows that long words generate greater activation that is available both earlier and for a longer time than is the case for short words. As a result, lexical influences on phoneme identification are extremely robust for long words but are quite fragile and condition-dependent for short words. Models of word recognition must consider words of all lengths to capture the true dynamics of lexical activation. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Atomistic simulations of grain boundary migration in copper

While the motion of twist boundaries can be readily studied by atomistic simulations with molecular dynamics (MD) under the action of an elastic driving force, the approach fails for tilt boundaries. This is due to the interaction of the elastic stress with the grain boundary (GB) structure, which causes plastic strain by GB sliding. A novel concept, the orientation correlated driving force, is introduced to circumvent this problem. It is shown that this concept can be successfully applied to the study of the migration of tilt boundaries. The migration behavior of several twist and tilt GBs was investigated. The transition from low-to high-angle boundaries can be captured, and a structural transition of tilt boundaries was found at high temperatures, which also affected the migration behavior. The results compare well with experimental results of the motion high-angle boundaries, but for low-angle boundaries, the agreement is poor. This article is based on a presentation made in the “Hillert Symposium on Thermodynamics & Kinetics of Migrating Interfaces in Steels and Other Complex Alloys,” December 2–3, 2004, organized by The Royal Institute of Technology in Stockholm, Sweden.     

On the credibility of manet simulations

Simulation is useful for evaluating protocol performance and operation. However, the lack of rigor with which it's applied threatens the credibility of the published research within the manet research community. Mobile ad hoc networks (manets) allow rapid deployment because they don't depend on a fixed infrastructure. Manet nodes can participate as the source, the destination, or an intermediate router. This flexibility is attractive for military applications, disaster-response situations, and academic environments where fixed net working infrastructures might not be available.     

Reliable eye-diagram analysis of data links via device macromodels

This paper addresses the impact of device macromodels on the accuracy of signal integrity and performance predictions for critical digital interconnecting systems. It exploits nonlinear parametric models for both single-ended and differential devices, including the effects of power supply fluctuations and receiver bit detection. The analysis demonstrates that the use of well-designed macromodels dramatically speeds up the simulation as well it preserves timing accuracy even for long bit sequences.     

On-line estimation of kinetic parameters in anaerobic digestion using observer-based estimators and multiwavelength fluorometry.

Observer-based estimators (OBE) were used for estimation of state variables and kinetic parameters in an anaerobic digestion (AD) process. A simplified first-order model with time-varying kinetic parameters was used to design an OBE for kinetic parameter estimation. This approach was validated on a laboratory-scale anaerobic reactor equipped with a multiwavelength fluorometer for on-line measurements of COD and VFA concentrations in the reactor effluent. The proposed estimators provide continuous adjustment of kinetic parameters and can be used for predictions of state variables between samples acquisition and during sensor failure.     

Identification of the adverse effect of nitrate on the phosphate release rate and improvement of EBPR process models.

The adverse effect of nitrate on the phosphate release rate in the anaerobic phase was observed and was hardly explainable with conventional EBPR process models. Four possible mechanisms were proposed including substrate competition, reduced fermentation, parallel reaction and sequential reaction. Batch experiments were designed and conducted to identify the dominant mechanism. Results showed that the sequential reaction was the only possible mechanism where only denitrification occurred if any nitrate existed in the anaerobic phase. Then the phosphate release following after the nitrate was completely removed. Nitrate inhibition effect was added into the PHA storage rate to incorporate the sequential reaction in the conventional ASM3 plus EAWAG bio-P module (ASM3 + P). Nitrate inhibition coefficient, K(I,NO,PAO) was found to be as low as 0.05 mg/L. This correlated well with experimental observation where no also meant that the anaerobic compartment of a continuous flow reactor could be seriously affected by the residual nitrate contained in the sludge recycle flow. This phenomenon caused overestimation of the phosphate uptake rate and consequently underestimation of PO4(3-) -P concentration. This problem was resolved by incorporation of a nitrate inhibition term in the ASM3 + P for more accurate simulation of the EBPR process.