---------------------------------------------------------------------------|
Deep learning framework for geological symbol detection on geological maps
Please unzip the zip files and read the following tutorial.
How to use the datasets:
Datasets used for the experiments articulated in the paper have been
provided and located in paths ./SSDN_DATASET and ./DAGCN_DATASET.
Please unzip the related dataset zip files and modify parameters of the
dataset root paths set in ./SSDN/train.py, ./DAGCN/train_DAGCN.py, and
./DAGCN/train_Other_GCNN.py according to your device. Then, the code can be
run directly.
Note that, using DAGCN_DATASET/DatasetProcess/Dizhitu_graph/Process/
data.zip->Dizhitu_graph_2_for_Paper2_1_added_EXP_Chg_1 as an example, the
documents named in this format are referred to the experiments mentioned in
Section 5.2, where `Chg` is the compound symbol name and the `1` denotes
the number of the `Chg` existing in the modified dataset.
Please use `labelme` (a site package of python) to visualize the raw
datasets (see ./SSDN_DATASET_RAW and ./DAGCN_DATASET_RAW). The interpretation
of the marks for `DAGCN_DATASET_RAW` is located in the APPENDIX B of this file.
How to use the codes:
The codes in this repository are all python codes and can be run by
Pycharm directly. Please see the following guide for further operations
and correct the path(s) according to your own device(s), when you run any
python files.
Single Symbol Detection Network:
Run ./SSDN/train.py to train the model. Run ./SSDN/train_see_results
.py to see the training result. Please modify the class Config in the
above-mentioned files in accordance with your purpose.
Distance-Attention Graph Convolutional Network:
Run ./DAGCN/train_DAGCN.py to train the model. Run ./DAGCN/train_Other
_GCNN.py to train other GCN models. Training and testing results will be
recorded automatically by local files. Please modify the parameters in the
related python files to control the network structure.
Thank you for your reading!
---------------------------------------------------------------------------|
---------------------------------------------------------------------------|
APPENDIX A: PYTHON ENVIRONMENT
python 3.6.3
APPENDIX PYTHON Site-packages
albumentations 0.4.3
apex 0.1
Cython 0.29.14
labelme 4.5.7
matplotlib 3.3.1
numpy 1.18.1
opencv-contrib-python 4.1.2.30
opencv-python-headless 4.1.2.30
scikit-image 0.16.2
torch 1.1.0
torch-geometric 1.4.2
torch-scatter 1.2.0
torch-sparse 0.4.0
torchvision 0.2.1
---------------------------------------------------------------------------|
---------------------------------------------------------------------------|
APPENDIX B:
{
'0': "background_class",
"a01": "ArFai",
"a10": "ArHggv",
"a1": "Ar2SggyDeltaO",
"a12": "Ar2SHggyDeltaO",
"a13": "Ar2WggvDelta",
"a20": "Ar3FN",
"a201": "Ar3CN",
"a2": "Ar3Ggg",
"a21": "Ar3JggyDelta",
"a22": "Ar3Kggyo",
"a23": "Ar3Qgg",
"Ce1": "Cent1+2",
"5": "Cent2m+zh",
"6": "Cent3chm",
"c": "Cent1ch",
"c00": "Ch",
"c1": "Chg",
"c2": "Chd",
"c21": "Chd+g",
"c30": "Chch",
"c30t": "Chch-t ",
"c31": "Chchl-d",
"c32": "Chchl+t",
"c3": "Chch+chl",
"c4": "Cht+d",
"c41": "Cht",
"E": "Exl",
"7": "Jxw",
"J": "Jxh",
"t": "Jxt",
"j2": "J2j",
"j21": "J2l",
"y": "J3t",
"y1": "J3tch",
"2": "J3shh",
"f0": "J3dsh",
"f": "J3fsh",
"f1": "J3shy",
"g1": "J3lch",
"g2": "J3tp",
"g21": "J3xj",
"g22": "Jxy",
"g3": "J3au",
"g31": "J3EfecilongPai",
"g4": "J3zh",
"1": "K1xsh",
"3": "K1hsh",
"31": "K1shch",
"9": "K1dj",
"k1": "K1jx",
"k2": "K1tl",
"k3": "K1ym",
"n1": "Nql",
"pt2": "Pt2Shd",
"pt3": "Pt2dy",
"4": "Qbl+j",
"jey": "Qbj",
"lsz": "Qbl",
"8": "Qbx",
"q1": "Qhpl",
"q11": "Qhpal",
"q12": "Qhal",
"qp2": "Qp2Zh",
"qpm1": "Qp3mpal",
"qpm2": "Qp3mpl",
"qpm3": "Qp3m",
"qpm4": "Qp3mdl",
"GuJ": "EBetaMiu",
"Tq1": "TqbDeltaMiu",
"Tz1": "TzhwyPai",
}
---------------------------------------------------------------------------|