scholarly journals AN EFFICIENT METHOD TO DETECT MUTUAL OVERLAP OF A LARGE SET OF UNORDERED IMAGES FOR STRUCTURE-FROM-MOTION

2017 ◽  
Vol IV-1/W1 ◽  
pp. 191-198 ◽  
Author(s):  
X. Wang ◽  
Z.Q. Zhan ◽  
C. Heipke
Keyword(s):  
Structure From Motion ◽  
Low Cost ◽  
Feature Space ◽  
New Method ◽  
Large Set ◽  
General Applicability ◽  
Nearest Neighbours ◽  
Speed Up ◽  
Common Block ◽  
Overlap Graphs

Recently, low-cost 3D reconstruction based on images has become a popular focus of photogrammetry and computer vision research. Methods which can handle an arbitrary geometric setup of a large number of unordered and convergent images are of particular interest. However, determining the mutual overlap poses a considerable challenge.<br><br> We propose a new method which was inspired by and improves upon methods employing random k-d forests for this task. Specifically, we first derive features from the images and then a random k-d forest is used to find the nearest neighbours in feature space. Subsequently, the degree of similarity between individual images, the image overlaps and thus images belonging to a common block are calculated as input to a structure-from-motion (sfm) pipeline. In our experiments we show the general applicability of the new method and compare it with other methods by analyzing the time efficiency. Orientations and 3D reconstructions were successfully conducted with our overlap graphs by sfm. The results show a speed-up of a factor of 80 compared to conventional pairwise matching, and of 8 and 2 compared to the VocMatch approach using 1 and 4 CPU, respectively.

scholarly journals TRIMMING FEATURE EXTRACTION AND INFERENCE FOR MCU-BASED EDGE NILM: A SYSTEMATIC APPROACH

2021 ◽  
Author(s):  
Enrico Tabanelli ◽  
Davide Brunelli ◽  
Luca Benini ◽  
Andrea Acquaviva
Keyword(s):  
Cost Reduction ◽  
State Of The Art ◽  
Low Cost ◽  
Feature Space ◽  
Low Latency ◽  
Voltage Sensors ◽  
Global Power ◽  
Feature Spaces ◽  
Load Monitoring ◽  
Speed Up

Non-Intrusive Load Monitoring (NILM) enables the disaggregation of the global power consumption of multiple loads, taken from a single smart electrical meter, into appliance-level details. State-of-the-Art approaches are based on Machine Learning methods and exploit the fusion of time- and frequency-domain features from current and voltage sensors. Unfortunately, these methods are compute-demanding and memory-intensive. Therefore, running low-latency NILM on low-cost, resource-constrained MCU-based meters is currently an open challenge. This paper addresses the optimization of the feature spaces as well as the computational and storage cost reduction needed for executing State-of-the-Art (SoA) NILM algorithms on memory- and compute-limited MCUs. We compare four supervised learning techniques on different classification scenarios and characterize the overall NILM pipeline’s implementation on a MCU-based Smart Measurement Node. Experimental results demonstrate that optimizing the feature space enables edge MCU-based NILM with 95.15% accuracy, resulting in a small drop compared to the most-accurate feature vector deployment (96.19%) while achieving up to 5.45x speed-up and 80.56% storage reduction. Furthermore, we show that low-latency NILM relying only on current measurements reaches almost 80% accuracy, allowing a major cost reduction by removing voltage sensors from the hardware design.

scholarly journals TRIMMING FEATURE EXTRACTION AND INFERENCE FOR MCU-BASED EDGE NILM: A SYSTEMATIC APPROACH

2021 ◽  
Author(s):  
Enrico Tabanelli ◽  
Davide Brunelli ◽  
Luca Benini ◽  
Andrea Acquaviva
Keyword(s):  
Cost Reduction ◽  
State Of The Art ◽  
Low Cost ◽  
Feature Space ◽  
Low Latency ◽  
Voltage Sensors ◽  
Global Power ◽  
Feature Spaces ◽  
Load Monitoring ◽  
Speed Up

Non-Intrusive Load Monitoring (NILM) enables the disaggregation of the global power consumption of multiple loads, taken from a single smart electrical meter, into appliance-level details. State-of-the-Art approaches are based on Machine Learning methods and exploit the fusion of time- and frequency-domain features from current and voltage sensors. Unfortunately, these methods are compute-demanding and memory-intensive. Therefore, running low-latency NILM on low-cost, resource-constrained MCU-based meters is currently an open challenge. This paper addresses the optimization of the feature spaces as well as the computational and storage cost reduction needed for executing State-of-the-Art (SoA) NILM algorithms on memory- and compute-limited MCUs. We compare four supervised learning techniques on different classification scenarios and characterize the overall NILM pipeline’s implementation on a MCU-based Smart Measurement Node. Experimental results demonstrate that optimizing the feature space enables edge MCU-based NILM with 95.15% accuracy, resulting in a small drop compared to the most-accurate feature vector deployment (96.19%) while achieving up to 5.45x speed-up and 80.56% storage reduction. Furthermore, we show that low-latency NILM relying only on current measurements reaches almost 80% accuracy, allowing a major cost reduction by removing voltage sensors from the hardware design.

scholarly journals A New Method for the Large-Scale Documentation of Pottery Sherds Through Simultaneous Multiple 3D Model Capture Using Structure from Motion: Phoenician Carinated-Shoulder Amphorae from Tell el-Burak (Lebanon) as a Case Study

2021 ◽  
Vol 7 (1) ◽  
pp. 256-272
Author(s):  
Fanet Göttlich ◽  
Aaron Schmitt ◽  
Andrea Kilian ◽  
Helen Gries ◽  
Kamal Badreshany
Keyword(s):  
Structure From Motion ◽  
Large Scale ◽  
3D Model ◽  
Low Cost ◽  
Cost Effective ◽  
New Method ◽  
Large Collection ◽  
Subsequent Research ◽  
Pottery Sherds

Abstract This paper presents a new rapid, low-cost method for the large-scale documentation of pottery sherds through simultaneous multiple 3D model capture using Structure from Motion (SfM). The method has great potential to enhance and replace time-consuming and expensive conventional approaches for pottery documentation, i.e., 2D photographs and drawing on paper with subsequent digitization of the drawings. To showcase the method’s effectiveness and applicability, a case study was developed in the context of an investigation of the Phoenician economy at the Lebanese site of Tell el-Burak, which is based on a large collection of amphora sherds. The same set of sherds were drawn by an experienced draftsperson and then documented through SfM using our new workflow to allow for a direct comparison. The results show that the new technique detailed here is accessible, more cost-effective, and allows for the documentation of ceramic data at a far-greater scale, while producing more consistent and reproducible results. We expect that these factors will enable excavators to greatly increase digital access to their material, which will significantly enhance its utility for subsequent research.

Rapid mapping accuracy assessment using low-cost UAV and open-source Structure from Motion

2020 ◽  
Vol 52 ◽  
pp. 55-61
Author(s):  
Ettore Potente ◽  
Cosimo Cagnazzo ◽  
Alessandro Deodati ◽  
Giuseppe Mastronuzzi
Keyword(s):  
Open Source ◽  
Structure From Motion ◽  
Accuracy Assessment ◽  
Low Cost ◽  
Mapping Accuracy ◽  
Rapid Mapping ◽  
Source Structure

scholarly journals Experiments of Image Classification Using Dissimilarity Spaces Built with Siamese Networks

Sensors ◽  
2021 ◽  
Vol 21 (5) ◽  
pp. 1573
Author(s):  
Loris Nanni ◽  
Giovanni Minchio ◽  
Sheryl Brahnam ◽  
Gianluca Maguolo ◽  
Alessandra Lumini
Keyword(s):  
Vector Space ◽  
Image Classification ◽  
Ad Hoc ◽  
Feature Space ◽  
Medical Data ◽  
Training Data ◽  
Data Sets ◽  
Large Set ◽  
Clustering Methods ◽  
Siamese Networks

Traditionally, classifiers are trained to predict patterns within a feature space. The image classification system presented here trains classifiers to predict patterns within a vector space by combining the dissimilarity spaces generated by a large set of Siamese Neural Networks (SNNs). A set of centroids from the patterns in the training data sets is calculated with supervised k-means clustering. The centroids are used to generate the dissimilarity space via the Siamese networks. The vector space descriptors are extracted by projecting patterns onto the similarity spaces, and SVMs classify an image by its dissimilarity vector. The versatility of the proposed approach in image classification is demonstrated by evaluating the system on different types of images across two domains: two medical data sets and two animal audio data sets with vocalizations represented as images (spectrograms). Results show that the proposed system’s performance competes competitively against the best-performing methods in the literature, obtaining state-of-the-art performance on one of the medical data sets, and does so without ad-hoc optimization of the clustering methods on the tested data sets.

scholarly journals Clustering-Based Component Fraction Estimation in Solid–Liquid Two-Phase Flow in Dredging Engineering

Sensors ◽  
2020 ◽  
Vol 20 (19) ◽  
pp. 5697
Author(s):  
Chang Sun ◽  
Shihong Yue ◽  
Qi Li ◽  
Huaxiang Wang
Keyword(s):  
Clustering Algorithm ◽  
Two Phase Flow ◽  
Low Cost ◽  
Fast Response ◽  
New Method ◽  
Phase Flow ◽  
Reference Distribution ◽  
Two Phase ◽  
Long Time ◽  
Solid Liquid

Component fraction (CF) is one of the most important parameters in multiple-phase flow. Due to the complexity of the solid–liquid two-phase flow, the CF estimation remains unsolved both in scientific research and industrial application for a long time. Electrical resistance tomography (ERT) is an advanced type of conductivity detection technique due to its low-cost, fast-response, non-invasive, and non-radiation characteristics. However, when the existing ERT method is used to measure the CF value in solid–liquid two-phase flow in dredging engineering, there are at least three problems: (1) the dependence of reference distribution whose CF value is zero; (2) the size of the detected objects may be too small to be found by ERT; and (3) there is no efficient way to estimate the effect of artifacts in ERT. In this paper, we proposed a method based on the clustering technique, where a fast-fuzzy clustering algorithm is used to partition the ERT image to three clusters that respond to liquid, solid phases, and their mixtures and artifacts, respectively. The clustering algorithm does not need any reference distribution in the CF estimation. In the case of small solid objects or artifacts, the CF value remains effectively computed by prior information. To validate the new method, a group of typical CF estimations in dredging engineering were implemented. Results show that the new method can effectively overcome the limitations of the existing method, and can provide a practical and more accurate way for CF estimation.

Breeding for improved blanchability in peanut: phenotyping, genotype × environment interaction and selection

2018 ◽  
Vol 69 (12) ◽  
pp. 1237 ◽  
Author(s):  
G. C. Wright ◽  
M. G. Borgognone ◽  
D. J. O Connor ◽  
R. C. N. Rachaputi ◽  
R. J. Henry ◽  
...  
Keyword(s):  
Low Cost ◽  
Peanut Butter ◽  
Environment Interaction ◽  
Breeding Lines ◽  
Seed Supply ◽  
Genotype Environment Interaction ◽  
Effective Selection ◽  
The Usa ◽  
Speed Up ◽  
Very High

Breeding for improved blanchability—the propensity of the testa (skin) to be removed from the kernel following rapid heat treatment—is a priority for improvement in the Australian Peanut Breeding Program (APBP). Easy removal of the testa by blanching is required for processing of peanuts into peanut butter and various other confectionary products. Thus, blanchability is an economically important trait in any newly released cultivar in Australia. A better understanding of the range of genetic variation, nature of inheritance and genotype×environment (G×E) interactions, and the development of a low-cost method to phenotype in early generations, could speed up breeding for this trait. Studies were conducted to develop a low-cost, rapid method utilising minimal amounts of seed to phenotype in early generations, along with an assessment of G×E interactions over a range of years and environments to derive optimal selection protocols. Use of a smaller kernel sample size than standard (50 vs 200g) was effective for accurately assessing blanchability in breeding lines and could allow selection in early generations (e.g. in seed produced from a single F2 plant where seed supply is adequate). G×E interaction for blanchability was shown to be very low. Genotypic variance explained 62–100% of the total variance for blanchability, assessed in two diverse germplasm pools including 107 accessions in the USA mini-core over three environments and multiple APBP breeding lines grown over nine different years–environments. Genotypic correlations between all environments were very high (~0.60–0.96), with heritability for the blanchability trait estimated to be very high (0.74–0.97) across the 13 trials. The results clearly demonstrate that effective selection for improved blanchability can be conducted in early generations and in a limited number of contrasting environments to ensure consistency of results.

scholarly journals Brief Communication: Mapping river ice using drones and structure from motion

2018 ◽  
Vol 12 (2) ◽  
pp. 627-633 ◽  
Author(s):  
Knut Alfredsen ◽  
Christian Haas ◽  
Jeffrey A. Tuhtan ◽  
Peggy Zinke
Keyword(s):  
Structure From Motion ◽  
Low Cost ◽  
Cold Climate ◽  
River Ice ◽  
River Morphology ◽  
Ice Thickness ◽  
Surface Model ◽  
Engineering Data ◽  
Site Access ◽  
Remote Sensing Methods

Abstract. In cold climate regions, the formation and break-up of river ice is important for river morphology, winter water supply, and riparian and instream ecology as well as for hydraulic engineering. Data on river ice is therefore significant, both to understand river ice processes directly and to assess ice effects on other systems. Ice measurement is complicated due to difficult site access, the inherent complexity of ice formations, and the potential danger involved in carrying out on-ice measurements. Remote sensing methods are therefore highly useful, and data from satellite-based sensors and, increasingly, aerial and terrestrial imagery are currently applied. Access to low cost drone systems with quality cameras and structure from motion software opens up a new possibility for mapping complex ice formations. Through this method, a georeferenced surface model can be built and data on ice thickness, spatial distribution, and volume can be extracted without accessing the ice, and with considerably fewer measurement efforts compared to traditional surveying methods. A methodology applied to ice mapping is outlined here, and examples are shown of how to successfully derive quantitative data on ice processes.

scholarly journals Rapid method of screening for drought stress tolerance in maize (Zea mays L.)

2020 ◽  
Vol 80 (01) ◽  
Author(s):  
Bhupender Kumar ◽  
Krishan Kumar ◽  
Shankar Lal Jat ◽  
Shraddha Srivastava ◽  
Tanu Tiwari ◽  
...  
Keyword(s):  
Drought Stress ◽  
Drought Tolerance ◽  
Stress Tolerance ◽  
Low Cost ◽  
Large Set ◽  
Antioxidant Genes ◽  
Screening Technique ◽  
Drought Stress Tolerance ◽  
Hydroponic Solution ◽  
Drought Screening

Drought stress is the major production constraint in rainfed maize. Screening for drought tolerance is severely affected by the lack of a simple and reliable phenotyping technique. The objective of this study was to standardize a simple hydroponic based drought screening technique in maize. In this context, one week old uniform seedlings of 55 inbreds and 5 hybrids were transferred to hydroponic solution in the glass house. The seedlings were allowed to acclimatize for next one week in hydroponic solution. The drought stress was imposed by removing seedlings from nutrient solution and exposed to air for 6 and 4 hours daily for a period of 5 and 4 consecutive days in hybrids and inbreds, respectively. Data were recorded on all shoot and root parameters, and based on stress symptoms, a drought tolerance score was given to each genotype. The percent deductions in shoot and root fresh weight from non-stress to stress ranged from 11.7 to 84.4 and 2.1 to 77.5, respectively. Six inbred lines, namely, DQL790-4, CML334, CM140, CML422, CM125 and HKI488 and three hybrids namely DMRH1306, DMRH1410 and PMH4 were found drought tolerant. The effectiveness of this screening technique was compared and confirmed using pots screening as well as by expression profiling of key antioxidant genes (Sod2, Sod4, Sod9 and Apx1) playing role in drought stress tolerance. This phenotyping technique is very short, low cost and simple which can be utilized in preliminary drought screening for large set of maize germplasm and mapping populations.

A simple and low-cost method for leaf area measurement of detached leaves

1997 ◽  
Vol 33 (01) ◽  
pp. 65-72 ◽  
Author(s):  
J. T. Korva ◽  
G. A. Forbes
Keyword(s):  
Solanum Tuberosum ◽  
Leaf Area ◽  
Low Cost ◽  
Electronic Equipment ◽  
New Method ◽  
Area Measurement ◽  
Water Spray ◽  
Spray Method ◽  
Detached Leaves ◽  
Highly Correlated

A technique for leaf area measurement utilizing water spray as an inexpensive substitute for electronic equipment was developed and tested with leaves of potato (Solanum tuberosum L.). The leaf areas measured by the spray method were highly correlated with those measured by an electronic area meter. Measurements of leaf area obtained by the spray method were significantly more highly correlated with those obtained by the area meter than were the measurements of dry weights. The main advantages of the new method are precision, accuracy and immediate results at a low cost.

Sign in / Sign up

Export Citation Format

Share Document