NU-InNet: Thai Food Image Recognition Using Convolutional Neural Networks on Smartphone
원문 초록
Currently, Convolutional Neural Networks (CNN) have been widely used in many applications. Image recognition is one of the applications utilizing CNN. For most of the research in this field, CNN is used mainly to increase the effectiveness of the recognition. However, the processing time and the amount of the parameters (or model size) are not taken into account as the main factors. In this paper, the image recognition for Thai food using a smartphone is studied. The processing time and the model size are reduced so that they can be properly used with smartphones. A new network called NUInNet (Naresuan University Inception Network) that adopts the concept of Inception module used in GoogLeNet is proposed in the paper. It is applied and tested with Thai food database called THFOOD-50, which contains 50 kinds of famousThai food. It is found that NU-InNet can reduce the processing time and the model size by the factors of 2 and 10, respectively, comparing to those obtained from GoogLeNet while maintaining the recognition precision to the same level as GoogLeNet. This significant reduction in the processing time and the model size using the proposed network can certainly satisfy users for Thai-food recognition application in a smartphone.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
ChefFusion: Multimodal Foundation Model Integrating Recipe and Food Image Generation
Res-VMamba: Fine-Grained Food Category Visual Classification Using Selective State Space Models with Deep Residual Learning
Feature-Suppressed Contrast for Self-Supervised Food Pre-training
A Central Asian Food Dataset for Personalized Dietary Interventions, Extended Abstract
Vision and Structured-Language Pretraining for Cross-Modal Food Retrieval
Mining Discriminative Food Regions for Accurate Food Recognition
토론
이 논문과 연결된 포럼 글이 아직 없습니다.