{"id":27285233,"url":"https://github.com/snigdhap2301/floorverse","last_synced_at":"2026-05-20T07:33:48.658Z","repository":{"id":287400494,"uuid":"964610906","full_name":"snigdhap2301/FLOORVERSE","owner":"snigdhap2301","description":"A Universe of Floor Plan Possibilities","archived":false,"fork":false,"pushed_at":"2025-04-18T07:44:06.000Z","size":8664,"stargazers_count":1,"open_issues_count":0,"forks_count":0,"subscribers_count":1,"default_branch":"main","last_synced_at":"2025-04-18T21:42:15.782Z","etag":null,"topics":["architecture-project","conditional-rendering","cvae","dataset","deep-learning","efficiency-analysis","flask","floorplan-reconstruction","floorplanning","generative-adversarial-network","machine-learning-algorithms","python","residential-floorplan"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/snigdhap2301.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null}},"created_at":"2025-04-11T13:45:54.000Z","updated_at":"2025-04-18T07:44:09.000Z","dependencies_parsed_at":"2025-04-11T15:03:34.333Z","dependency_job_id":null,"html_url":"https://github.com/snigdhap2301/FLOORVERSE","commit_stats":null,"previous_names":["snigdhap2301/floorverse"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/snigdhap2301/FLOORVERSE","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/snigdhap2301%2FFLOORVERSE","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/snigdhap2301%2FFLOORVERSE/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/snigdhap2301%2FFLOORVERSE/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/snigdhap2301%2FFLOORVERSE/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/snigdhap2301","download_url":"https://codeload.github.com/snigdhap2301/FLOORVERSE/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/snigdhap2301%2FFLOORVERSE/sbom","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":266682625,"owners_count":23967837,"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","status":"online","status_checked_at":"2025-07-23T02:00:09.312Z","response_time":66,"last_error":null,"robots_txt_status":null,"robots_txt_updated_at":null,"robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"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":["architecture-project","conditional-rendering","cvae","dataset","deep-learning","efficiency-analysis","flask","floorplan-reconstruction","floorplanning","generative-adversarial-network","machine-learning-algorithms","python","residential-floorplan"],"created_at":"2025-04-11T19:22:45.595Z","updated_at":"2026-05-20T07:33:48.628Z","avatar_url":"https://github.com/snigdhap2301.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# FLOORVERSE: A Universe of Floor Plan Possibilities\n\nFLOORVERSE is an innovative platform that leverages cutting-edge AI to automate and optimize the generation of residential floorplans. By utilizing a *Conditional Variational Autoencoder (CVAE)* model, meticulously trained on the extensive **RPlan dataset** (comprising over 80,000 images), FLOORVERSE produces functionally coherent and aesthetically pleasing layouts based on user-defined constraints. The system seamlessly integrates a Flask-based backend with an interactive frontend, providing users with real-time, personalized design experiences.\n\n## Architecture\n\nAt the heart of FLOORVERSE is a **Conditional Variational Autoencoder (CVAE)**, designed to process visual data from floorplan images in conjunction with user-defined design parameters. The architecture is structured into the following key components:\n\n1.  **Encoder**:  Utilizes convolutional layers to efficiently extract spatial features directly from grayscale floorplan images.\n2.  **Reparameterization Trick**:  Employs this technique to sample latent vectors in a differentiable manner, crucial for effective backpropagation and model training.\n3.  **Decoder**:  Reconstructs existing floorplans or generates entirely new ones based on the learned latent representations and the specific user conditions provided.\n4.  **Training \u0026 Optimization**:  The model is trained using a composite loss function that combines binary cross-entropy for image reconstruction and KL divergence to ensure a well-structured latent space, leading to robust and reliable performance.\n\n## Workflow\n\nThe FLOORVERSE workflow follows a streamlined process:\n\n1.  Input Preprocessing: Raw floorplan images undergo preprocessing steps such as resizing and normalization to prepare them for model input.\n2.  Feature Encoding: The Encoder network processes the preprocessed visual data, while condition vectors representing user inputs are processed through a multi-layer perceptron to capture relevant design constraints.\n3.  Latent Space Representation: Fully connected layers are used to generate a compressed latent space representation that encapsulates the essence of the input floorplan and user conditions.\n4.  Floorplan Generation: The Decoder network takes the latent space representation and user constraints to reconstruct or generate new floorplans that are tailored to the specified requirements.\n\n## Evaluation and Metrics\n\nThe performance of FLOORVERSE has been rigorously evaluated through a combination of quantitative metrics and qualitative analyses to ensure both accuracy and design quality.\n\n### Quantitative Metrics\n\n*   *Reconstruction Loss*: 8571.3582 - Measures how well the model reconstructs input floorplans, indicating the fidelity of the encoder-decoder process.\n*   *Total Loss*: 8654.9813 - Represents the overall loss during training, combining reconstruction loss and KL divergence to optimize model performance.\n*   *KL Divergence*: 83.6231 - Quantifies the divergence between the learned latent space distribution and a standard normal distribution, ensuring a well-organized and continuous latent space for generation.\n\n### Qualitative Analyses\n\n*   *Latent Space Traversals*: Demonstrated the ability to smoothly interpolate between diverse design archetypes -  Showcases the model's capability to generate a spectrum of floorplan styles and configurations by navigating the learned latent space.\n*   *Principal Component Analysis (PCA)*: Highlighted spatial coherence in generated layouts -  Visualizes the latent space and confirms that generated layouts maintain spatial relationships and structural integrity, resulting in realistic and well-structured floorplans.\n\n\n\n## Usage\n\n### Clone the repository\n```bash\ngit clone https://github.com/your-repo/FLOORVERSE.git\ncd FLOORVERSE\n```\n### Prerequisites\n```bash\npip install -r requirements.txt\n```\n### Start the Flask server\n```bash\npython app.py\n```\n\n\n### Documentation\nComprehensive documentation for FLOORVERSE is available `documentation/` directory to help you understand the project in detail, from setup to advanced usage and development.\n\n## Contributors\n\nThis project was a collaborative effort by a team of dedicated individuals, including:\n\n- [Snigdha Pandey](https://github.com/snigdhap2301)\n- [Ansh Prakash](https://github.com/anshprakash6397)\n- [Esther George Sam](https://github.com/esthersam07)\n- [Munish Thakur](https://github.com/menotthakur)\n- [Naveen Singh ](https://github.com/naveenks2002)\n\n\n\n## Acknowledgments\nFLOORVERSE is built upon the foundation of extensive research in AI-driven architectural design. We gratefully acknowledge the insights and advancements from prior works in generative adversarial networks, graph neural networks, and diffusion-based approaches within the field.\nWe extend our sincere gratitude to the creators of the RPlan dataset for providing a robust and invaluable resource that enabled the training of our model.\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsnigdhap2301%2Ffloorverse","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fsnigdhap2301%2Ffloorverse","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fsnigdhap2301%2Ffloorverse/lists"}