Research:
Introduction
The proposed dataset was developed entirely for the purpose of meeting the requirements of the proposed model, and due to the lack of one that meets them. It consists of images of the shapes or objects, shadows and intensities, which for this research are a fundamental part and the essential resource to achieve the proposed objectives.
In addition, geometric objects have a common characteristic, which is that they are uniform in their edges as well as in their structure and shadows. The latter can be processed in different models, convolutional neural network architectures, with deep learning and using algorithms with which the experimentation process would be carried out, to determine the height and generate the volume of the object from the shadow it casts.
One of the motivations in proposing this research is the possibility that with the information that is collected, processed and analyzed, through the use of neural networks, a deep learning model and a set of data can be obtained, which can be used to find the right combination of parameters for the shape from the shadows and light.
Hence the importance of this topic because it is expected to prove that it is possible to combine deep learning, a data set, the analysis of light and shadows for the development of an algorithm that in experimentation obtains optimal results. This being a theoretical-practical reference and a starting point to issues related to shadows, understanding that in these there is more information than can be perceived with the naked eye.
Read More
