AWS-DeepLens: How to access the pre-trained model for object recognition in S3

Question

I am using AWS with the DeepLens device and I want to fine-tune a model for object-recognition. Because DeepLens still has limitations with the layers it can support I want to start from the same model that the demo project has.

According to the description of the demo project the model is at s3://deeplens-managed-resources/models/SSDresNet50.

However, when I open the S3 service there is no deeplens-managed-resources bucket.

Does anyone know why could that be and how I can access the model?

PS: If more information is needed please let me know so that I can edit my answer to include it.

Answer 1

I dont know about the S3 managed bucket but you can find the pretrained model inside your deeplens when you deploy the sample project. This model files can be obtained in /opt/awscam/artifacts/ location and then you can pick it to retrain your model.

Answer 2

I could not find a way to access the deeplens-managed-resources bucket from the AWS console on a web browser.

However, one can access it with the AWS-cli and use the "aws s3 sync" command to download the bucket's content in a local directory.