{"id":20401922,"url":"https://github.com/jmmanley/conv-autoencoder","last_synced_at":"2025-04-12T14:23:26.216Z","repository":{"id":49901917,"uuid":"153361297","full_name":"jmmanley/conv-autoencoder","owner":"jmmanley","description":"a convolutional autoencoder in python and keras.","archived":false,"fork":false,"pushed_at":"2020-05-25T14:59:24.000Z","size":77,"stargazers_count":5,"open_issues_count":1,"forks_count":5,"subscribers_count":2,"default_branch":"master","last_synced_at":"2024-04-26T21:47:49.596Z","etag":null,"topics":["autoencoder","computer-vision","convolutional-autoencoder","convolutional-neural-networks","keras","python"],"latest_commit_sha":null,"homepage":null,"language":"Jupyter Notebook","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"gpl-3.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/jmmanley.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2018-10-16T22:15:09.000Z","updated_at":"2022-07-27T11:51:50.000Z","dependencies_parsed_at":"2022-08-29T17:11:33.599Z","dependency_job_id":null,"html_url":"https://github.com/jmmanley/conv-autoencoder","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jmmanley%2Fconv-autoencoder","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jmmanley%2Fconv-autoencoder/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jmmanley%2Fconv-autoencoder/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/jmmanley%2Fconv-autoencoder/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/jmmanley","download_url":"https://codeload.github.com/jmmanley/conv-autoencoder/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248579019,"owners_count":21127746,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["autoencoder","computer-vision","convolutional-autoencoder","convolutional-neural-networks","keras","python"],"created_at":"2024-11-15T04:52:17.175Z","updated_at":"2025-04-12T14:23:26.158Z","avatar_url":"https://github.com/jmmanley.png","language":"Jupyter Notebook","funding_links":[],"categories":[],"sub_categories":[],"readme":"# Convolutional Autoencoder (CAE) in Python\n\nAn implementation of a convolutional autoencoder in python and keras.\n\n## Installation\n\n`pip install cae`\n\n## Usage\n\n```\nfrom cae import cae\nimport numpy as np\n\n# create a fake dataset, here: 1000 random 224x224 RGB images\nimages = np.random.normal(size=(1000, 224, 224, 3))\n\nlatent_dim = 8 # desired latent dimensionality\n\nmodel = cae(images.shape[1:], latent_dim) # there are a number of **kwargs\n                                          # parameters that are likely\n                                          # worth tuning!!!\n\n# TRAIN THE NETWORK\nmodel.fit(images)\n\n# SAVE THE WEIGHTS FOR EASY RELOADING LATER WITH model.load_weights(path)\nmodel.save_weights('PATH/TO/SAVE/')\n```\n\n## Final words\n\n`cae.py` contains the implementation, which is tested on the MNIST dataset in `mnist_test.ipynb`.\n\nIn general, auto-encoders map an input x to a latent representation y (generally in a much smaller dimensional space), using deterministic functions of the type y = sigma(Wx+b). In order to encode images, it is useful to implement a convolutional architecture. Here, we utilize convolutional layers and max-pooling layers (which allow translation-invariant representations), followed by a flattening and dense layer to encode the images in a reduced-dimensional space. For decoding, you essentially need to perform the inverse operation. For more information on CAEs, consult e.g. http://people.idsia.ch/~ciresan/data/icann2011.pdf.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjmmanley%2Fconv-autoencoder","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fjmmanley%2Fconv-autoencoder","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fjmmanley%2Fconv-autoencoder/lists"}