| Resumo: | The population growth in the last few decades has led to the development of urban areas, which induced an increased difficulty in finding quality food. The difficulty in finding quality nourishment and a growing offer in the fast-food industry due to the fast pace at which life is lived in big cities has caused increasing obesity and sedentary lifestyle. In 2016 more than 1.9 billion adults aged 18 years and older were overweight[1]. However, this tendency has started to reverse, and with the increasing concern for diseases such as obesity and diabetes, people started return to shopping in farmers mar kets and choosing wisely the locals where they eat, which led to the development of more healthy fast food chains. This new tendency has made new technologies appear that were created to help improve customer choices and facilitate choosing the best food items that have the best quality. This dissertation will analyse the different devices and solutions in the market, such as near-infrared sensors and computer vision. The objective of this dissertation is to build a system that can detect which type of food item we choose and obtain nutritional information. The development begins with researching the different options of small devices that already exist in the market and with which a person can take shopping and assist them by obtaining the nutritional information, such as SCIO or Tellspec. This device cannot detect which type of food is being analysed, so human interaction it is still needed to obtain the best results possible. However, it can return the nutritional information necessary for the first part of this dissertation’s development. Besides being small (palm-handed), these sensors are also cheap and faster compared to equivalent laboratory equipment. The second objective of this dissertation was developed to solve the lack of detection of which type of food is present in the module. To solve this problem and taking into account the objective, it was decided to use computer vision and, more specifically, image recognition and deep machine learning applied in food databases. This dissertation’s main objective is to create a module that can classify and obtain the nutritional information of different types of food. It also serves as a helping hand in the kitchen to control the quality and quantity of the food that the user ingests daily. There will be an exhaustive testing session for the near-infrared sensors using different types of fruits to prove the concept. For the computer vision, it will be applied a deep learning algorithm with supervised training to obtain a high accuracy result.
|
|---|