Skip to content

Files

Latest commit

AnkitMethwanineelamgehlot
AnkitMethwani
and
neelamgehlot
Update Sample Notebooks. (aws#872)
Sep 6, 2019
7dbfb55 · Sep 6, 2019

History

History
This branch is 1928 commits behind aws/amazon-sagemaker-examples:default.

imageclassification_caltech

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
Image-classification-fulltraining-elastic-inference.ipynb
Image-classification-fulltraining-elastic-inference.ipynb
Sep 6, 2019
Image-classification-fulltraining-highlevel.ipynb
Image-classification-fulltraining-highlevel.ipynb
May 16, 2019
Image-classification-fulltraining.ipynb
Image-classification-fulltraining.ipynb
Sep 6, 2019
Image-classification-incremental-training-highlevel.ipynb
Image-classification-incremental-training-highlevel.ipynb
Nov 7, 2018
Image-classification-lst-format-highlevel.ipynb
Image-classification-lst-format-highlevel.ipynb
Jul 15, 2019
Image-classification-lst-format.ipynb
Image-classification-lst-format.ipynb
Jul 15, 2019
Image-classification-transfer-learning-highlevel.ipynb
Image-classification-transfer-learning-highlevel.ipynb
Dec 8, 2018
Image-classification-transfer-learning.ipynb
Image-classification-transfer-learning.ipynb
Sep 6, 2019
README.md
README.md
Jul 22, 2019

README.md

SageMaker Image classification full training

This notebook ImageClassification-fulltraining.ipynb demos an end-2-end system for image classification training using resnet model. Caltech-256 dataset is used as a sample dataset. Various parameters such as network depth (number of layers), batch_size, learning_rate, etc., can be varied in the training. Once the training is complete, the notebook shows how to host the trained model for inference.

SageMaker Image classification transfer learning

This notebook Imageclassification-transfer-learning.ipynb demos an end-2-end system for image classification fine-tuning using a pre-trained resnet model on imagenet dataset. Caltech-256 dataset is used as a transfer learning dataset. The network re-initializes the output layer with the number of classes in the Caltech dataset and retrains the layer while at the same time fine-tuning the other layers. Various parameters such as network depth (number of layers), batch_size, learning_rate, etc., can be varied in the training. Once the training is complete, the notebook shows how to host the trained model for inference.

SageMaker Image classification lst format

This notebook Imageclassification-lst-format.ipynb demos an end-2-end system for image classification training with image and list files. Caltech-256 dataset is used as a transfer learning dataset. The network re-initializes the output layer with the number of classes in the Caltech dataset and retrains the layer while at the same time fine-tuning the other layers. Various parameters such as network depth (number of layers), batch_size, learning_rate, etc., can be varied in the training. Once the training is complete, the notebook shows how to host the trained model for inference.

SageMaker Image classification full training highlevel

This notebook ImageClassification-fulltraining-highlevel.ipynb is similar to the ImageClassification-fulltraining.ipynb but using Sagemaker high-level APIs

SageMaker Image classification transfer learning highlevel

This notebook Imageclassification-transfer-learning-highlevel.ipynb is similar to the ImageClassification-transfer-learning.ipynb but using Sagemaker high-level APIs

SageMaker Image classification lst format highlevel

This notebook Imageclassification-lst-format-highlevel.ipynb is similar to the ImageClassification-lst-format.ipynb but using Sagemaker high-level APIs