参考url

※1 Deep Floor Plan Recognition Using a Multi-Task Network With Room-Boundary-Guided Attention

python仮想環境の作成

$ pyenv install 3.7.16
$ pyenv install 3.10.10
$ pyenv versions
* system (set by /home/end0tknr/.pyenv/version)
* 3.7.16
  3.10.10
  3.11.2

$ pyenv local 3.10.10
$ python --version
Python 3.10.10

$ cd /home/end0tknr/proj
$ python -m venv deep_floor_plan
$ cd deep_floor_plan
$ source ./bin/activate

git cloneや pip install

GPUなしの環境を使用していますので、以下では pip install tensorflow-cpu しています

$ cd /home/end0tknr/proj/deep_floor_plan
$ git clone https://github.com/zcemycl/TF2DeepFloorplan.git

$ pip install gdown
$ pip install --upgrade --no-cache-dir gdown
$ pip install matplotlib numpy opencv-python pdbpp \
              tensorboard tensorflow-cpu scipy Pillow gdown

学習済みデータのdownload & 解凍

$ cd /home/end0tknr/proj/deep_floor_plan
$ gdown https://drive.google.com/uc?id=1czUSFvk6Z49H-zRikTc67g2HUUz4imON
$ unzip log.zip
$ rm log.zip

Deep Floor Plan for python の実行

python script

import tensorflow as tf
import sys
sys.path.append('./TF2DeepFloorplan/')
sys.path.append('./TF2DeepFloorplan/dfp')
from dfp.net import *
from dfp.data import *
import matplotlib.image as mpimg
import matplotlib.pyplot as plt
from argparse import Namespace
import os
import gc
os.environ['TF_FORCE_GPU_ALLOW_GROWTH'] = 'false'
sys.path.append('./TF2DeepFloorplan/dfp/utils/')
from dfp.utils.rgb_ind_convertor import *
from dfp.utils.util import *
from dfp.utils.legend import *
from dfp.deploy import *

#print( tf.config.list_physical_devices('GPU') )
#print(tf.test.is_gpu_available())

args = Namespace(image='./TF2DeepFloorplan/resources/30939153.jpg',
                 weight='./log/store/G',
                 loadmethod='log',
                 postprocess=True,
                 colorize=True,
                 save="output.png")
result = dfp.deploy.main(args)

# 凡例を legend.png として出力
# dfp.utils.legend.main()

標準出力結果

実行時にwarningのようなメッセージが表示されましたが、生成された画像には悪影響していないようでした。

2023-05-06 15:10:40.477830: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2, in other operations, rebuild TensorFlow with the appropriate compiler flags.
<module 'dfp.utils._paths' from '/home/end0tknr/proj/deep_floor_plan/./TF2DeepFloorplan/dfp/utils/_paths.py'>
<module 'dfp._paths' from '/home/end0tknr/proj/deep_floor_plan/./TF2DeepFloorplan/dfp/_paths.py'>
2023-05-06 15:10:44.888348: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 67108864 exceeds 10% of free system memory.
2023-05-06 15:10:44.937080: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 67108864 exceeds 10% of free system memory.
2023-05-06 15:10:44.954258: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 67108864 exceeds 10% of free system memory.
2023-05-06 15:10:44.970637: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 67108864 exceeds 10% of free system memory.
2023-05-06 15:10:45.170290: W tensorflow/tsl/framework/cpu_allocator_impl.cc:83] Allocation of 67108864 exceeds 10% of free system memory.
/home/end0tknr/proj/deep_floor_plan/./TF2DeepFloorplan/dfp/utils/util.py:33: UserWarning: ignoring size because footprint is set
  ndimage.grey_erosion(output_array, size=(3, 3), footprint=el),
(600, 419, 3) (600, 419, 3)