House-GAN++: Generative Adversarial Layout Refinement Network towards Intelligent Computational Agent for Professional Architects

Last update: Dec 28, 2022

Related tags

Deep Learning houseganpp

Overview

House-GAN++

Code and instructions for our paper: House-GAN++: Generative Adversarial Layout Refinement Network towards Intelligent Computational Agent for Professional Architects, CVPR 2021. Project website.

Data

We have used the RPLAN dataset, which offers 60k vector-graphics floorplans designed by professional architects. Qualitative and quantitative evaluations based on the three standard metrics (i.e., realism, diversity, and compatibility) in the literature demonstrate that the proposed system outperforms the current-state-of-the-art by a large margin.

Demo

Please check out our live demo.

Running pretrained models

See requirements.txt for checking the dependencies before running the code

For running a pretrained model check out the following steps:

Run python test.py.
Check out the results in output folder.

Training models

Coming Soon!

Citation

@misc{nauata2021housegan,
      title={House-GAN++: Generative Adversarial Layout Refinement Networks}, 
      author={Nelson Nauata and Sepidehsadat Hosseini and Kai-Hung Chang and Hang Chu and Chin-Yi Cheng and Yasutaka Furukawa},
      year={2021},
      eprint={2103.02574},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}

Contact

If you have any question, feel free to contact me at [email protected]

Acknowledgement

This research is partially supported by NSERC Discovery Grants, NSERC Discovery Grants Accelerator Supplements, DND/NSERC Discovery Grant Supplement, and Autodesk. We would like to thank architects and students for participating in our user study.

Comments

Could you share the post-processing code for getting project demo-like (vector floorplan) results?

First of all, thanks for the amazing research. I am super interested in this design generation research.

I would like to get cleaner-looking results to see. In your paper, I found you got a pair-wise comparison with raw segmentation and also vector floorplan.

I would like to check the results from vector floorplan visualization.

If it is in your github repo already, could you tell me where it is? I didn't see a description of it.

And I have a question about the floorplan result. Once I got the testing result from test.py with "data/json" using pre-trained weights, I found that some of the results are missing the entrance door. (filename: 7513.json, 18477.json, 19307.json, 36233.json, 45012.json, 45161,json)

Do you know the reason? I suspect entrance doors are overlapped by other masks.

opened by sucream1004 5
ModuleNotFoundError: No module named 'pygraphviz'

Traceback (most recent call last): File "test.py", line 11, in from dataset.floorplan_dataset_maps_functional_high_res import FloorplanGraphDataset, floorplan_collate_fn File "/data/houseganpp/dataset/floorplan_dataset_maps_functional_high_res.py", line 28, in from misc.utils import ROOM_CLASS, ID_COLOR File "/data/houseganpp/misc/utils.py", line 30, in from pygraphviz import * ModuleNotFoundError: No module named 'pygraphviz'

opened by Ha0Tang 0
what's the reason input vector is 18-d ?

Hi, I noticed that the code using input vector is 18-d ,instead of 12-d vector in the paper. So, I am so curious why you did that, and what benefit that can make. thanks for any reply

opened by f745311 0
Dataloader error

Hi,

I am trying to reproduce the results of your paper and I am getting an error while loading data. The error is the following:

113 ./data/rplan/45776.json [Epoch 0/200] [Batch 111/379] [D loss: 9.999077] [G loss: 81.515854] [L1 loss: 83.728096] Traceback (most recent call last): File "train.py", line 114, in for i, batch in enumerate(fp_loader): File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 530, in next data = self._next_data() File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 1224, in _next_data 114 ./data/rplan/16164.json return self._process_data(data) File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 1250, in _process_data data.reraise() File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/_utils.py", line 457, in reraise raise exception IndexError: Caught IndexError in DataLoader worker process 0. Original Traceback (most recent call last): File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/utils/data/_utils/worker.py", line 287, in _worker_loop data = fetcher.fetch(index) File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py", line 49, in fetch data = [self.dataset[idx] for idx in possibly_batched_index] File "/home/user/miniconda3/envs/houseganpp/lib/python3.7/site-packages/torch/utils/data/_utils/fetch.py", line 49, in data = [self.dataset[idx] for idx in possibly_batched_index] File "/home/user/houseganpp/dataset/floorplan_dataset_maps_functional_high_res.py", line 135, in getitem graph_nodes, graph_edges, rooms_mks = self.build_graph(rms_type, fp_eds, eds_to_rms) File "/home/user/houseganpp/dataset/floorplan_dataset_maps_functional_high_res.py", line 294, in build_graph poly = self.make_sequence(np.array([fp_eds[l][:4] for l in eds_poly]))[0] File "/home/user/houseganpp/dataset/floorplan_dataset_maps_functional_high_res.py", line 189, in make_sequence v_curr = tuple(edges[0][:2]) IndexError: index 0 is out of bounds for axis 0 with size 0

I have built the database with the repository https://github.com/sepidsh/Housegan-data-reader

I attach the file with the error, but I think that the error is not in the file but in the dataloader. When I feed just this file to the model, it works. 16164.zip

Have you had this error before? Do you know how to solve it?

opened by deeplearningdiaries 1
how to process the RPLAN dataset
I use the code in misc/read_floorplan.py to process the RPLAN dataset,

but I find the bboxes is not computed，and always [] !!!!!

So, how to get the bboxes

bboxes, edges, ed_rm = [], [], []
info=dict() for w_i in range(len(walls)): edges.append([((walls[w_i][0]-min_x)/lenx),((walls[w_i][1]-min_y)/leny),((walls[w_i][2]-min_x)/lenx),((walls[w_i][3]-min_y)/leny),walls[w_i][5],walls[w_i][8]]) if(walls[w_i][6]==-1): ed_rm.append([walls[w_i][7]]) elif(walls[w_i][7]==-1): ed_rm.append([walls[w_i][6]]) else: ed_rm.append([walls[w_i][6],walls[w_i][7]])

for i in range(len(poly)): p=poly[i] pm=[] for p_i in range((p)): if(p_i%2==0): pm.append(([edges[km+p_i][0],edges[km+p_i][1]])) else: pm.append(([edges[km+p_i][2],edges[km+p_i][3]])) km=km+p info['room_type'] = room_type info['boxes'] = bboxes info['edges'] = edges info['ed_rm'] = ed_rm
opened by czhxiaohuihui 1
System Exit : 2

Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True). usage: ipykernel_launcher.py [-h] [--n_cpu N_CPU] [--batch_size BATCH_SIZE] [--checkpoint CHECKPOINT] [--data_path DATA_PATH] [--out OUT] ipykernel_launcher.py: error: unrecognized arguments: -f /root/.local/share/jupyter/runtime/kernel-2b829f88-b0ca-4d50-b119-ca22aa9df5a4.json An exception has occurred, use %tb to see the full traceback.

SystemExit: 2

opened by maryam-ghaderi-bafti 1

House-GAN++: Generative Adversarial Layout Refinement Network towards Intelligent Computational Agent for Professional Architects

Related tags

Overview

House-GAN++

Data

Demo

Running pretrained models

Training models

Citation

Contact

Acknowledgement

Comments

Could you share the post-processing code for getting project demo-like (vector floorplan) results?

ModuleNotFoundError: No module named 'pygraphviz'

what's the reason input vector is 18-d ?

Dataloader error

how to process the RPLAN dataset

System Exit : 2

Owner

A method that utilized Generative Adversarial Network (GAN) to interpret the black-box deep image classifier models by PyTorch.

CityLearn Challenge Multi-Agent Reinforcement Learning for Intelligent Energy Management, 2020, PikaPika team

A platform for intelligent agent learning based on a 3D open-world FPS game developed by Inspir.AI.

This code uses generative adversarial networks to generate diverse task allocation plans for Multi-agent teams.

A computational optimization project towards the goal of gerrymandering the results of a hypothetical election in the UK.

PyTorch 1.5 implementation for paper DECOR-GAN: 3D Shape Detailization by Conditional Refinement.

NR-GAN: Noise Robust Generative Adversarial Networks

HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis

π-GAN: Periodic Implicit Generative Adversarial Networks for 3D-Aware Image Synthesis

Partial implementation of ODE-GAN technique from the paper Training Generative Adversarial Networks by Solving Ordinary Differential Equations

Flickr-Faces-HQ (FFHQ) is a high-quality image dataset of human faces, originally created as a benchmark for generative adversarial networks (GAN)

Generate high quality pictures. GAN. Generative Adversarial Networks

This project uses reinforcement learning on stock market and agent tries to learn trading. The goal is to check if the agent can learn to read tape. The project is dedicated to hero in life great Jesse Livermore.

Hierarchical-Bayesian-Defense - Towards Adversarial Robustness of Bayesian Neural Network through Hierarchical Variational Inference (Openreview)

Official PyTorch implementation of the paper "Recycling Discriminator: Towards Opinion-Unaware Image Quality Assessment Using Wasserstein GAN", accepted to ACM MM 2021 BNI Track.

CVPR2021: Temporal Context Aggregation Network for Temporal Action Proposal Refinement

RefineGNN - Iterative refinement graph neural network for antibody sequence-structure co-design (RefineGNN)

Speech Enhancement Generative Adversarial Network Based on Asymmetric AutoEncoder

pytorch implementation for Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network arXiv:1609.04802