scholarly journals Task-Agnostic Object Recognition for Mobile Robots through Few-Shot Image Matching

Electronics ◽  
2020 ◽  
Vol 9 (3) ◽  
pp. 380 ◽  
Author(s):  
Agnese Chiatti ◽  
Gianluca Bardaro ◽  
Emanuele Bastianelli ◽  
Ilaria Tiddi ◽  
Prasenjit Mitra ◽  
...  
Keyword(s):  
Object Recognition ◽  
Mobile Robots ◽  
Image Matching ◽  
Scale Up ◽  
Target Object ◽  
Data Set ◽  
Vast Number ◽  
Training Time ◽  
Novel Objects ◽  
Object Classes

To assist humans with their daily tasks, mobile robots are expected to navigate complex and dynamic environments, presenting unpredictable combinations of known and unknown objects. Most state-of-the-art object recognition methods are unsuitable for this scenario because they require that: (i) all target object classes are known beforehand, and (ii) a vast number of training examples is provided for each class. This evidence calls for novel methods to handle unknown object classes, for which fewer images are initially available (few-shot recognition). One way of tackling the problem is learning how to match novel objects to their most similar supporting example. Here, we compare different (shallow and deep) approaches to few-shot image matching on a novel data set, consisting of 2D views of common object types drawn from a combination of ShapeNet and Google. First, we assess if the similarity of objects learned from a combination of ShapeNet and Google can scale up to new object classes, i.e., categories unseen at training time. Furthermore, we show how normalising the learned embeddings can impact the generalisation abilities of the tested methods, in the context of two novel configurations: (i) where the weights of a Convolutional two-branch Network are imprinted and (ii) where the embeddings of a Convolutional Siamese Network are L2-normalised.

scholarly journals Affine Shape Descriptor Algorithm for Image Matching and Object Recognition in a Digital Image/Algoritmo para clasificación de formas invariante a transformaciones afines para objetos rígidos en una imagen digital

2017 ◽  
Vol 6 (11) ◽  
pp. 13
Author(s):  
Jacob Morales G. ◽  
Nancy Arana D. ◽  
Alberto A. Gallegos
Keyword(s):  
Comparative Analysis ◽  
Object Recognition ◽  
Digital Image ◽  
Image Matching ◽  
State Of The Art ◽  
Shape Descriptor ◽  
Correct Classification ◽  
Affine Transformations ◽  
Affine Shape ◽  
Object Classes

The use of shape, as a mean to discriminate between object classes extracted from a digital image, is one of the major roles in machine vision. The use of shape has been studied extensively in recent decades, because the shape of the object holds enough information for its correct classification; additionally, the quantity of memory used to store a border is much less than that of the whole region within it. In this paper, a novel shape descriptor is proposed. The algorithm demonstrates that it has useful properties such as: invariance to affine transformations that are applied to the border (e.g., scales, skews, displacements and rotations), stability in the presence of noise, and good differentiability between different object classes. A comparative analysis is included to show the performance of our proposal with respect to the state of the art algorithms.

Some statistical and CI models to predict chaotic high-frequency financial data

2020 ◽  
Vol 39 (5) ◽  
pp. 6419-6430
Author(s):  
Dusan Marcek
Keyword(s):  
Time Series Data ◽  
Moving Average ◽  
Methodological Approach ◽  
Back Propagation ◽  
Large Data ◽  
Series Data ◽  
Data Set ◽  
Training Time ◽  
Optimal Population ◽  
Forecast Time

To forecast time series data, two methodological frameworks of statistical and computational intelligence modelling are considered. The statistical methodological approach is based on the theory of invertible ARIMA (Auto-Regressive Integrated Moving Average) models with Maximum Likelihood (ML) estimating method. As a competitive tool to statistical forecasting models, we use the popular classic neural network (NN) of perceptron type. To train NN, the Back-Propagation (BP) algorithm and heuristics like genetic and micro-genetic algorithm (GA and MGA) are implemented on the large data set. A comparative analysis of selected learning methods is performed and evaluated. From performed experiments we find that the optimal population size will likely be 20 with the lowest training time from all NN trained by the evolutionary algorithms, while the prediction accuracy level is lesser, but still acceptable by managers.

Multi‐image matching for object recognition

2018 ◽  
Vol 12 (3) ◽  
pp. 350-356 ◽  
Author(s):  
Jie Zhu ◽  
Shufang Wu ◽  
Xizhao Wang ◽  
Guoqing Yang ◽  
Liyan Ma
Keyword(s):  
Object Recognition ◽  
Image Matching

scholarly journals Forecast skill score assessment of a relocatable ocean prediction system, using a simplified objective analysis method

Ocean Science ◽  
2017 ◽  
Vol 13 (6) ◽  
pp. 925-945 ◽  
Author(s):  
Reiner Onken
Keyword(s):  
Observational Data ◽  
Skill Score ◽  
Forecast Skill ◽  
Objective Analysis ◽  
Prediction System ◽  
Regional Ocean Modeling System ◽  
Data Set ◽  
Vast Number ◽  
Ocean Prediction ◽  
Forecast Skill Score

Abstract. A relocatable ocean prediction system (ROPS) was employed to an observational data set which was collected in June 2014 in the waters to the west of Sardinia (western Mediterranean) in the framework of the REP14-MED experiment. The observational data, comprising more than 6000 temperature and salinity profiles from a fleet of underwater gliders and shipborne probes, were assimilated in the Regional Ocean Modeling System (ROMS), which is the heart of ROPS, and verified against independent observations from ScanFish tows by means of the forecast skill score as defined by Murphy(1993). A simplified objective analysis (OA) method was utilised for assimilation, taking account of only those profiles which were located within a predetermined time window W. As a result of a sensitivity study, the highest skill score was obtained for a correlation length scale C = 12.5 km, W = 24 h, and r = 1, where r is the ratio between the error of the observations and the background error, both for temperature and salinity. Additional ROPS runs showed that (i) the skill score of assimilation runs was mostly higher than the score of a control run without assimilation, (i) the skill score increased with increasing forecast range, and (iii) the skill score for temperature was higher than the score for salinity in the majority of cases. Further on, it is demonstrated that the vast number of observations can be managed by the applied OA method without data reduction, enabling timely operational forecasts even on a commercially available personal computer or a laptop.

scholarly journals A New Paradigm for Training Hyperactive Dopamine Transporter Knockout Rats: Influence of Novel Stimuli on Object Recognition

2021 ◽  
Vol 15 ◽  
Author(s):  
Natalia P. Kurzina ◽  
Anna B. Volnova ◽  
Irina Y. Aristova ◽  
Raul R. Gainetdinov
Keyword(s):  
Object Recognition ◽  
Dopamine Transporter ◽  
Cognitive Task ◽  
Recognition Task ◽  
Long Term Memory ◽  
New Paradigm ◽  
Dopaminergic Transmission ◽  
Novel Objects ◽  
High Level ◽  
The Brain

Attention deficit hyperactivity disorder (ADHD) is believed to be connected with a high level of hyperactivity caused by alterations of the control of dopaminergic transmission in the brain. The strain of hyperdopaminergic dopamine transporter knockout (DAT-KO) rats represents an optimal model for investigating ADHD-related pathological mechanisms. The goal of this work was to study the influence of the overactivated dopamine system in the brain on a motor cognitive task fulfillment. The DAT-KO rats were trained to learn an object recognition task and store it in long-term memory. We found that DAT-KO rats can learn to move an object and retrieve food from the rewarded familiar objects and not to move the non-rewarded novel objects. However, we observed that the time of task performance and the distances traveled were significantly increased in DAT-KO rats in comparison with wild-type controls. Both groups of rats explored the novel objects longer than the familiar cubes. However, unlike controls, DAT-KO rats explored novel objects significantly longer and with fewer errors, since they preferred not to move the non-rewarded novel objects. After a 3 months’ interval that followed the training period, they were able to retain the learned skills in memory and to efficiently retrieve them. The data obtained indicate that DAT-KO rats have a deficiency in learning the cognitive task, but their hyperactivity does not prevent the ability to learn a non-spatial cognitive task under the presentation of novel stimuli. The longer exploration of novel objects during training may ensure persistent learning of the task paradigm. These findings may serve as a base for developing new ADHD learning paradigms.

scholarly journals Twitter, Telepractice, and the COVID-19 Pandemic: A Social Media Content Analysis

2021 ◽  
pp. 1-11
Author(s):  
Kristen Weidner ◽  
Joneen Lowman ◽  
Anne Fleischer ◽  
Kyle Kosik ◽  
Peyton Goodbread ◽  
...  
Keyword(s):  
Social Media ◽  
Content Analysis ◽  
Technology Implementation ◽  
Scale Up ◽  
Media Content Analysis ◽  
Data Set ◽  
Implementation Framework ◽  
Health Crisis ◽  
Wide Scale ◽  
Twitter Users

Purpose Telepractice was extensively utilized during the COVID-19 pandemic. Little is known about issues experienced during the wide-scale rollout of a service delivery model that was novel to many. Social media research is a way to unobtrusively analyze public communication, including during a health crisis. We investigated the characteristics of tweets about telepractice through the lens of an established health technology implementation framework. Results can help guide efforts to support and sustain telehealth beyond the pandemic context. Method We retrieved a historical Twitter data set containing tweets about telepractice from the early months of the pandemic. Tweets were analyzed using a concurrent mixed-methods content analysis design informed by the nonadoption, abandonment, scale-up, spread, and sustainability (NASSS) framework. Results Approximately 2,200 Twitter posts were retrieved, and 820 original tweets were analyzed qualitatively. Volume of tweets about telepractice increased in the early months of the pandemic. The largest group of Twitter users tweeting about telepractice was a group of clinical professionals. Tweet content reflected many, but not all, domains of the NASSS framework. Conclusions Twitter posting about telepractice increased during the pandemic. Although many tweets represented topics expected in technology implementation, some represented phenomena were potentially unique to speech-language pathology. Certain technology implementation topics, notably sustainability, were not found in the data. Implications for future telepractice implementation and further research are discussed.

scholarly journals On solving large p-median problems

2019 ◽  
Vol 47 (6) ◽  
pp. 981-996
Author(s):  
Wangshu Mu ◽  
Daoqin Tong
Keyword(s):  
Big Data ◽  
Random Sampling ◽  
Computing Time ◽  
Scale Up ◽  
Problem Solution ◽  
Median Problem ◽  
Data Set ◽  
Location Models ◽  
Spatial Voting ◽  
Median Problems

Incorporating big data in urban planning has great potential for better modeling of urban dynamics and more efficiently allocating limited resources. However, big data may present new challenges for problem solutions. This research focuses on the p-median problem, one of the most widely used location models in urban and regional planning. Similar to many other location models, the p-median problem is non-deterministic polynomial-time hard (NP-hard), and solving large-sized p-median problems is difficult. This research proposes a high performance computing-based algorithm, random sampling and spatial voting, to solve large-sized p-median problems. Instead of solving a large p-median problem directly, a random sampling scheme is introduced to create smaller sub- p-median problems that can be solved in parallel efficiently. A spatial voting strategy is designed to evaluate the candidate facility sites for inclusion in obtaining the final problem solution. Tests with the Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) data set show that random sampling and spatial voting provides high-quality solutions and reduces computing time significantly. Tests also demonstrate the dynamic scalability of the algorithm; it can start with a small amount of computing resources and scale up and down flexibly depending on the availability of the computing resources.

scholarly journals Stereo Vision-Based Object Recognition and Manipulation by Regions with Convolutional Neural Network

Electronics ◽  
2020 ◽  
Vol 9 (2) ◽  
pp. 210 ◽  
Author(s):  
Yi-Chun Du ◽  
Muslikhin Muslikhin ◽  
Tsung-Han Hsieh ◽  
Ming-Shyan Wang
Keyword(s):  
Neural Network ◽  
Object Recognition ◽  
Convolutional Neural Network ◽  
Stereo Vision ◽  
Fuzzy Inference ◽  
Target Object ◽  
Robot Arm ◽  
Segmentation Method ◽  
Inference System ◽  
Six Degree Of Freedom

This paper develops a hybrid algorithm of adaptive network-based fuzzy inference system (ANFIS) and regions with convolutional neural network (R-CNN) for stereo vision-based object recognition and manipulation. The stereo camera at an eye-to-hand configuration firstly captures the image of the target object. Then, the shape, features, and centroid of the object are estimated. Similar pixels are segmented by the image segmentation method, and similar regions are merged through selective search. The eye-to-hand calibration is based on ANFIS to reduce computing burden. A six-degree-of-freedom (6-DOF) robot arm with a gripper will conduct experiments to demonstrate the effectiveness of the proposed system.

Improve Language Modeling for Code Completion Through Learning General Token Repetition of Source Code with Optimized Memory

2019 ◽  
Vol 29 (11n12) ◽  
pp. 1801-1818
Author(s):  
Yixiao Yang ◽  
Xiang Chen ◽  
Jiaguang Sun
Keyword(s):  
Prediction Accuracy ◽  
Language Model ◽  
Source Code ◽  
Attention Mechanism ◽  
Data Set ◽  
Training Time ◽  
Unseen Data ◽  
Code Completion ◽  
High Chance ◽  
Better Than

In last few years, applying language model to source code is the state-of-the-art method for solving the problem of code completion. However, compared with natural language, code has more obvious repetition characteristics. For example, a variable can be used many times in the following code. Variables in source code have a high chance to be repetitive. Cloned code and templates, also have the property of token repetition. Capturing the token repetition of source code is important. In different projects, variables or types are usually named differently. This means that a model trained in a finite data set will encounter a lot of unseen variables or types in another data set. How to model the semantics of the unseen data and how to predict the unseen data based on the patterns of token repetition are two challenges in code completion. Hence, in this paper, token repetition is modelled as a graph, we propose a novel REP model which is based on deep neural graph network to learn the code toke repetition. The REP model is to identify the edge connections of a graph to recognize the token repetition. For predicting the token repetition of token [Formula: see text], the information of all the previous tokens needs to be considered. We use memory neural network (MNN) to model the semantics of each distinct token to make the framework of REP model more targeted. The experiments indicate that the REP model performs better than LSTM model. Compared with Attention-Pointer network, we also discover that the attention mechanism does not work in all situations. The proposed REP model could achieve similar or slightly better prediction accuracy compared to Attention-Pointer network and consume less training time. We also find other attention mechanism which could further improve the prediction accuracy.

Sign in / Sign up

Export Citation Format

Share Document