INTERNATIONAL JOURNAL OF SCIENTIFIC DEVELOPMENT AND RESEARCH International Peer Reviewed & Refereed Journals, Open Access Journal ISSN Approved Journal No: 2455-2631 | Impact factor: 8.15 | ESTD Year: 2016
open access , Peer-reviewed, and Refereed Journals, Impact factor 8.15
In this paper, an efficient and fast approach has been proposed to localize every clearly visible food item from an image. For object detection we have processed every input image thoroughly to eliminate the complexities and the main limitations such as noise in the background, poor resolution and overlap between multiple objects in the image. We have implemented YOLO for object detection as it deals with object detection in a different possible way. It scans the entire image in a single instance or look and predicts the bounding box coordinates and class probabilities for these classified boxes. The advantage of using YOLO is its excellent speed – it’s incredibly fast and can process 45 frames per second. It also understands generalized object representation. Using this technique, we have developed a multiple food detection and recognition model regardless of various limitations. This model is also capable of detecting every single food item separately from a set of overlapping objects quite fast and another major advantage of our model is capable of detecting food items not only limited to the images from the dataset but is quite capable of detecting real world images. We also have proposed a model that determines the daily calorie requirements of an individual and also a BMI calculator that tells the current state of our body such as obese, underweight or healthy. The ultimate goal is to help the individual to make healthy food choices to remain healthy and fit.
"Healthy Diet System by Food Recognition Using Image Processing ", International Journal of Science & Engineering Development Research (www.ijsdr.org), ISSN:2455-2631, Vol.5, Issue 5, page no.505 - 508, May-2020, Available :http://www.ijsdr.org/papers/IJSDR2005080.pdf
Downloads:
000337071
Publication Details:
Published Paper ID: IJSDR2005080
Registration ID:191817
Published In: Volume 5 Issue 5, May-2020
DOI (Digital Object Identifier):
Page No: 505 - 508
Publisher: IJSDR | www.ijsdr.org
ISSN Number: 2455-2631
Facebook Twitter Instagram LinkedIn