finetune_alexnet_with_tensorflow

Why we included this project

Finetuning AlexNet on your own image dataset is the whole point of this repo, and the code is small enough to read end to end. You get the full AlexNet graph definition, a finetune script that takes plain text files of image paths and class labels, and a data generator wrapper for TensorFlow's input pipeline, plus a notebook for checking the pretrained weights against ImageNet images. That makes it a solid teaching example for how transfer learning is wired up, even if the TensorFlow API and the architecture itself are older than what teams would pick today.