A multi-model image recognizer

[ArthurGoshtasby]

A multi-model recognizer has been designed and implemented in Python. The recognizer has a user-friendly interface, implemented in PyQt6, to guide a user through various features of the package. As models. a fully-connected neural network, a convolutional neural network, a fine-tuned VGG16, a fine-tuned MobileNet, a fine-tuned DenseNet121, and a fine-tuned ResNet50 are used. Other models may be added to the package for increased recognition accuracy.

Currently, the package uses images of U.S. coins as input and the multi-model recognizer is trained to detect and recognize U.S. coins in images. To detect objects other than U.S. coins in an image, a segmentation method that can detect objects of interest in images should be added to the package.

The code is fully commented and should be easy to follow but, feel free to contact the author for ambiguities, questions, or suggestions. A.A.G.

---

This discussion was created from the release A multi-model image recognizer.