Dimensional analysis of objects in a 2D image
Anil Singh Parihar, Mayank Gupta, Vayu Sikka, Gurnaaz Kaur · 2017
The volume calculation and weight estimation of objects is used in many real world applications, such as calculating the nutritional content in food items, in forensic science, etc. While there are several existing models on volume estimation, a majority of these models are based on 3 dimensional images, which are difficult to obtain in real time and require expensive equipment. In this paper, a method is introduced to estimate the dimensions of an object in a 2D image provided by the user. Three tasks are performed-object recognition using the algorithm of Haar Cascades, dimension calculation using a reference object and a Conversion Factor, and volume estimation using conventional 3D shape models for regular objects and a prediction model based on regression analysis for human body. The observed values are compared with the ground truth and obtained an accuracy of 93.69%.