1,720,996 research outputs found
An original framework for understanding human actions and body language by using deep neural networks
Full text linkThe evolution of both fields of Computer Vision (CV) and Artificial Neural Networks (ANNs) has allowed the development of efficient automatic systems for the analysis of people's behaviour.
By studying hand movements it is possible to recognize gestures, often used by people to communicate information in a non-verbal way.
These gestures can also be used to control or interact with devices without physically touching them. In particular, sign language and semaphoric hand gestures are the two foremost areas of interest due to their importance in Human-Human Communication (HHC) and Human-Computer Interaction (HCI), respectively.
While the processing of body movements play a key role in the action recognition and affective computing fields. The former is essential to understand how people act in an environment, while the latter tries to interpret people's emotions based on their poses and movements;
both are essential tasks in many computer vision applications, including event recognition, and video surveillance.
In this Ph.D. thesis, an original framework for understanding Actions and body language is presented. The framework is composed of three main modules: in the first one, a Long Short Term Memory Recurrent Neural Networks (LSTM-RNNs) based method for the Recognition of Sign Language and Semaphoric Hand Gestures is proposed; the second module presents a solution based on 2D skeleton and two-branch stacked LSTM-RNNs for action recognition in video sequences; finally, in the last module, a solution for basic non-acted emotion recognition by using 3D skeleton and Deep Neural Networks (DNNs) is provided.
The performances of RNN-LSTMs are explored in depth, due to their ability to model the long term contextual information of temporal sequences, making them suitable for analysing body movements.
All the modules were tested by using challenging datasets, well known in the state of the art, showing remarkable results compared to the current literature methods
Hedonic Analysis of Dried Pasta Prices Using E-Commerce Data—An Explorative Study
Full text linkItaly is the world leader in the production of pasta and the Italian market is characterized by strong price competition among large industrial producers. Thus, recently, many small and medium firms have started to differentiate their products as a way to achieve higher margins and escape from price competition. Using data on the prices and characteristics of dried pasta sold online in the Italian market and a hedonic price model, we estimated the implicit prices associated with several attributes that are currently available for dried pasta. We find that the "artisanal" statement on pasta labeling is associated with the highest price premium. Also, results show that protected geographical indication, Halal and Kosher certifications, and the use of ancient wheat varieties are valuable features of dried pasta sold in the Italian market. Instead, a positive, albeit limited in magnitude, price premium is associated with dried pasta made using 100% Italian durum wheat semolina, the organic method, enriched with additional ingredients. Findings suggest that producers can differentiate their products by mostly emphasizing their small-scale production methods, the territorial connotation, and the cultural and environmental sustainability of production. Otherwise, certifying dried pasta as Halal or Kosher can represent a complementary or alternative strategy to differentiate the product and achieve a higher price
An empirical framework to study food labelling fraud: an application to the Italian extra-virgin olive oil market
No full textThe mislabelling of agricultural and food products is one of the most common types of food fraud. Despite the frequency with which labelling fraud occurs, there is no empirical framework to study its welfare implications, the probability that it may occur, and the measures that can limit its occurrence. We present an empirical framework to study the economic consequences of food labelling fraud in a differentiated products food market. Such framework requires the availability of sales data and the use of an ‘attribute-space' demand model. The model is applied to the Italian extra-virgin olive oil market to simulate the occurrence of fraudulent ‘100 per cent Italian' claims. Our results indicate that potential consumer losses due to overpayments for a false claim are higher than manufacturer gains, suggesting that labelling fraud results in welfare losses and not just in welfare transfers. Simulation results indicate that the level of the current administrative fines is not likely to be effective to discourage ‘100 per cent Italian' labelling fraud. Imposing larger fines or other measures negatively affecting a firm's image could be more effective in deterring labelling fraud
Intention to purchase active and intelligent packaging to reduce household food waste: Evidence from italian consumers
No full textInnovations in food packaging, such as active and intelligent ones, improve food safety and lower household food waste by extending product shelf life and providing information about food quality, respectively. The consumer adoption of such innovations could contribute to reaching one of the Sustainable Development Goals which calls for halving the per capita global food waste by 2030. Thus, this paper aims to investigate the consumers’ willingness to purchase active and intelligent packaging to reduce household food waste using a sample of 260 Italian consumers and a modified Theory of Planned Behavior (TPB) model. Using a structural equation model, findings show that respondents are more willing to purchase intelligent packaging rather than active packaging to reduce their wastes at home. Finally, attitudes, perceived behavioral control, awareness, and planning routines are the most important drivers of the intention to reduce household food waste
Intention to purchase milk packaged in biodegradable packaging: Evidence from italian consumers
No full textThe dairy industry generates large volumes of liquid waste that can be used to produce biopolymers, potentially employable for the creation of milk biodegradable bottles. In that regard, this paper aims to explore the consumers’ intention to purchase sustainable packages, as well as to assess the willingness to pay for it considering renewable packages made using organic waste feedstocks from the dairy industry (e.g., whey) and plant-based material (e.g., corn, sugarcane, etc.). To reach the stated objectives, we collected individual-level information (e.g., age, gender, education, income) from a convenient sample of 260 Italian consumers and a modified version of the Theory of Planned Behavior estimated using a structural equation model. Findings show that attitudes and perceived behavioral control are the most important drivers of the consumers’ intention to purchase sustainable packages. Finally, statistics show that respondents slightly prefer to purchase products packaged using plant-based biodegradable material, as well as most of the respondents show a low willingness to pay for milk offered in biodegradable packaging, regardless of the raw material used. Then, policymakers and companies should invest in educational/informational campaigns pointing out the beneficial effects on the environment from the purchase of foods in sustainable packaging. This may potentially increase the consumers’ intention to purchase, as well as their willingness to pay for plant-based and dairy whey-based packages by increasing the sustainability of the dairy supply chain
Digitalization within food supply chains to prevent food waste. Drivers, barriers and collaboration practices
No full textFood supply chains increasingly rely on big-data management solutions to foster collaboration across the food supply chain and improve business performance. However, little is known about collaboration practices that actors on the digital food supply chain adopt to solve problems such as food waste, or about the drivers and barriers related to the digital transformation of the food supply chain. Most of food waste studies rely on quantitative analysis, which cannot reveal relevant details about the tensions and dynamics of collaboration. We conducted a qualitative study drawing on eighteen in-depth interviews - of managers of large multinational and local organizations covering different and relevant roles on the digital food supply chain - to investigate how organizational and food supply chain processes are affected by the digitalization of the operations along the food supply chain. By triangulating emerging findings with literature on supply chain management we discuss different views about collaborative practices for food waste prevention in the food supply chain and provide insights on how supply chain design and firms' operations have been re-conceptualized with the usage of digital technologies and on the institutional forces both limiting (barriers) and fostering (drivers) the diffusion of the digital food supply chain
The Use of Agro-Food Chain By-Products and Foods of Plant Origin to Obtain High-Value-Added Foods
No full textThe increased consumer demand for sustainable, health-promoting foods has propelled research into plant-based products and the valorization of food by-products [...
- …
