Food Image Recognition via Superpixel Based Low-Level and Mid-Level Distance Coding for Smart Home Applications

Abstract

Abstract:Food image recognition is a key enabler for many smart home applications such as smart kitchen and smart personal nutrition log. In order to improve living experience and life quality, smart home systems collect valuable insights of users' preferences, nutrition intake and health conditions via accurate and robust food image recognition. In addition, efficiency is also a major concern since many smart home applications are deployed on mobile devices where high-end GPUs are not available. In this paper, we investigate compact and efficient food image recognition methods, namely low-level and mid-level approaches. Considering the real application scenario where only limited and noisy data are available, we first proposed a superpixel based Linear Distance Coding (LDC) framework where distinctive low-level food image features are extracted to improve performance. On a challenging small food image dataset where only 12 training images are available per category, our framework has shown superior performance in both accuracy and robustness. In addition, to better model deformable food part distribution, we extend LDC's feature-to-class distance idea and propose a mid-level superpixel food parts-to-class distance mining framework. The proposed framework show superior performance on a benchmark food image datasets compared to other low-level and mid-level approaches in the literature.

Smartphone apps for tracking food consumption and recommendations: Evaluating artificial intelligence-based functionalities, features and quality of current apps

“…Novel food recognition systems for dietary evaluation have been enabled by the increasing use of smartphones and the advancement of artificial intelligence and computer vision technologies. Various studies have already been conducted in this domain (Zheng et al, 2017;He et al, 2015;Ravì et al, 2015;Nayak et al, 2020;Chopra & Purwar, 2021). The first thing users need to do in food consumption tracking and recommendation apps is track their food consumption.…”

Food Recipe Ingredient Substitution Ontology Design Pattern

“…Admittedly, her sister has suggested that erythritol is a popular sugar substitute in recent times. However, two doubts remained to be resolved: (1) will erythritol not lose its sweetness during baking, (2) in what ratio to replace sugar with erythritol to get a similar sweetness?…”

“…The computational support for handling tasks, such as recommending ingredient substitutes, falls into a relatively new area of food computing [1]with a broad spectrum of applications, such as food recognition [2], detecting food intake with the use of sensors [3], and computational diet management [4]. Many recent approaches in food computing and diet management use artificial intelligence (AI), both statistical methods (based on various forms of machine learning) and symbolic ones (based, for instance, on the usage of knowledge graphs [5]).…”

Mid‐level deep Food Part mining for food image recognition

“…However, the performance is poor due to the low complexity of the methods. The mid-level methods, with much higher complexity and better performance, result in around 1-2 s/ image of speed on CPU implementation (0.8 s/image on 8 core CPU for RFDC [9] and 1.5 s/image on a single CPU for SFP [10]). On the other hand, benefiting from recently developed graphics processing unit (GPU) computing hardwares, deep learning methods achieved high speed on GPU despite the large complexity of deep learning models.…”

“…Bossard et al [9] proposed a random forest‐based method to mine FPs with high probability for each class. Zheng et al [10] introduced a density ratio‐based approach to mine FPs with more appearances in target class and less appearances in background classes. Since these models employ low‐level local features with IFV encoding to represent mid‐level FPs, their recognition power is limited.…”