JJEM: Special Edition 2 (December 2024)
JJEM: Special Edition 2 (December 2024)
2024-12-08
Detection of logo and package in Indian Food products
Jeevesh Kumar J, Manjunath H T
In our rapidly developing industrial society, brand recognition and packaging have become crucial role in driving consumer pattern. We know that logos are typically designed from texts, images, or both text and images. As part of this methodology here are the few things considered in logo detection including whether the logo is printed on the package and to determine the package name. We know that the logos in package is very important any brands to sustain its position and quality in the packing field. The development of the new approach includes the employment of algorithms for image processing and for improved identification of logos and to identify package name. However, primarily, it applies the Convolutional Neural Networks (CNNs) and additionally, it is employed techniques of transfer learning with the progeny of the base models. Also, the object detection frameworks such as ResNet50 is also incorporated into the system. These methods use region proposal network, multiscale feature maps, and anchor boxes to enhance the logo detection process among the images and make it more efficient. Advantage of using Faster RCNNs is that it allows the software to learn features in a hierarchical manner and learn logos on other conditions of images, or transfer learning, which allows it to detect logos with higher accuracy compared to other available methods.
Convolutional Neural Network, faster R-CNN, transfer learning