search

brownvc / fast-synth

Public code release for our CVPR 2019 paper "Fast and Flexible Indoor Scene Synthesis via Deep Convolutional Generative Models"
Other
54 stars 20 forks source link

Instructions on creating the dataset and training #5

Open fbhdr opened 2 years ago

fbhdr commented 2 years ago

Hi, I wonder how the dataset is generated. From my understanding, this repo only provides some tools for dataset generation but lacks a main script for calling the tools, something like create_data.py in deep-synth. Is my understanding correct?

I know SUNCG is not publicly available now but I have downloaded the dataset. In addition, I find it possible to run fast-synth with some existing available datasets and I am able to turn some existing datasets to SUNCG format. Since your codes are not related to SUNCG license, can you provide instructions and codes on generating the dataset?

It is also appreciated if you could provide instructions on running the codes, e.g. the training/testing commands/flags you use in the original paper.

Thanks!

kwang-ether commented 2 years ago

Did you send me an email regarding the same issue or was that from a different person?

July-egg commented 2 years ago

Hi, I wonder how the dataset is generated. From my understanding, this repo only provides some tools for dataset generation but lacks a main script for calling the tools, something like create_data.py in deep-synth. Is my understanding correct?

I know SUNCG is not publicly available now but I have downloaded the dataset. In addition, I find it possible to run fast-synth with some existing available datasets and I am able to turn some existing datasets to SUNCG format. Since your codes are not related to SUNCG license, can you provide instructions and codes on generating the dataset?

It is also appreciated if you could provide instructions on running the codes, e.g. the training/testing commands/flags you use in the original paper.

Thanks!

Hello! Could you please send me the dataset through email? I'll be very very thankful if possible!

KD-kAnEsHi commented 1 week ago

Hi, I wonder how the dataset is generated. From my understanding, this repo only provides some tools for dataset generation but lacks a main script for calling the tools, something like create_data.py in deep-synth. Is my understanding correct?

I know SUNCG is not publicly available now but I have downloaded the dataset. In addition, I find it possible to run fast-synth with some existing available datasets and I am able to turn some existing datasets to SUNCG format. Since your codes are not related to SUNCG license, can you provide instructions and codes on generating the dataset?

It is also appreciated if you could provide instructions on running the codes, e.g. the training/testing commands/flags you use in the original paper.

Thanks!

Hey! I apologize for the inconvenience, is there any way I could please get access to the dataset?