Applying deep learning for food image analysis
Visualitza/Obre
Estadístiques de LA Referencia / Recolecta
Inclou dades d'ús des de 2022
Cita com:
hdl:2117/192500
Tutor / directorRadeva, Petia
Tipus de documentProjecte Final de Màster Oficial
Data2020-04
Condicions d'accésAccés obert
Tots els drets reservats. Aquesta obra està protegida pels drets de propietat intel·lectual i
industrial corresponents. Sense perjudici de les exempcions legals existents, queda prohibida la seva
reproducció, distribució, comunicació pública o transformació sense l'autorització del titular dels drets
Abstract
With the increasing availability of data on the internet, deep learning techniques
have been on the rise these past decade. Food images specifically are one of
the most commonly shared types of image on social media. Because of this, the
problem of food image analysis has been receiving increasing attention these past
few years. However, it presents a series of challenges compared to other computer
vision problems, which has limited the progress on the field. Nevertheless, some
specific methods who capitalize on these challenges have been able to obtain
good results.
In this thesis, the method of multi-scale multi-view feature aggregation (MSMVFA)
applied to food recognition is explored. It is a strategy that has been able to
obtain state-of-the-art performances on the literature recently. It capitalizes on
merging information from different scales as well as different types, for instance
ingredient and dish features. By using data coming from different granularity
levels a more robust and discriminative classification is possible.
A detailed validation is provided to test the results of the method under different
conditions. We see that both multi-scale and multi-view aspects of the strategy
can be beneficial for the classification in various conditions for the food recognition
problem.
ii
TitulacióMÀSTER UNIVERSITARI EN INTEL·LIGÈNCIA ARTIFICIAL (Pla 2017)
Col·leccions
Fitxers | Descripció | Mida | Format | Visualitza |
---|---|---|---|---|
148697.pdf | 3,342Mb | Visualitza/Obre |