A complete solution

Earlier in this chapter, we described how to get started with a baseline solution for an image classification competition. In this section, we show how an ensemble team involving rockstars, such as Abhishek (https://www.kaggle.com/abhishek) and Tanul (https://www.kaggle.com/tanulsingh077), achieved a silver medal zone solution (36^th ) using a cleverly structured application of the components discussed above. The post summarizing their solution can be found here: https://www.kaggle.com/competitions/cassava-leaf-disease-classification/discussion/220628.

Their final solution was a combination of three models: EfficientNet-B7, EfficientNet-B3a, and SE-ResNext50. The models followed the pipeline described in the notebook: https://www.kaggle.com/code/abhishek/tez-faster-and-easier-training-for-leaf-detection/. A notable thing about this notebook is that it utilizes tez: a PyTorch trainer developed by Abhishek (https://github.com/abhishekkrthakur/tez). The idea behind...