search

ViktoriiaBaib / curvedataextraction

Automated curve data extraction tool
Other
7 stars 2 forks source link

readme

Installation

Automated curve data extraction. Works with colored curves. Cross-platform. Color decomposition with k-means. X units parser. Automatic axes scale recognition.

  1. Download and unzip folder from https://drive.google.com/file/d/1e0UTKwhgJN9DuD2qYsLcWcKd6WomvRkl/view?usp=sharing

  2. Download scripts above.

Part I: install packages and TF2 object detection API

  1. Create anaconda variable 
    conda create -n imgrec python=3.9
conda install -n imgrec pip
conda activate imgrec
  2. Install Tensorflow 2.7 
    pip install TensorFlow==2.7
  3. Install protobuf (probably it is already there after previous step; try installation via brew for macOS) 
    pip install protobuf
  4. Install some packages 
    pip install easyocr
pip install opencv-python==4.5.4.60
  5. Install TF object detection API

Create a directory for Tensorflow Models repository. Go to this directory and do

git clone https://github.com/tensorflow/models.git
cd models/research
protoc object_detection/protos/*.proto --python_out=.
cp object_detection/packages/tf2/setup.py .
python -m pip install .
  1. Check installation 
    python object_detection/builders/model_builder_tf2_test.py

Part II: use pre-trained model

  1. cd object_detection
mkdir inference_graph

  2. copy the content of the inference_graph from folder to the corresponding directory

  3. mkdir training

    copy labelmap.pbtxt from folder to training/.

  4. mkdir images

    copy file 0.png from folder to images/.

  5. copy file label_map_util_v2.py from folder to utils/.

  6. copy script object_detection_axes_legend.py from above (step 2) to models/research/object_detection/.

  7. run this file from models/research/object_detection/ as

    python object_detection_axes_legend.py

    You should obtain json and png files for Legend, X axis and Y axis in images/

Part III: run the rest of the project

  1. copy scripts posterization.py and color_decomposition.py from above (step 2) to models/research/object_detection/.

  2. Run file color_decomposition.py as

    python color_decomposition.py

    You should obtain png of color cut and palette as well as json and png of separate clusters in images/

  3. Copy scripts final-record.py and final_record_func.py from above (step 2) to models/research/object_detection/.

  4. Run final-record.py as

    python final-record.py

    You should obtain json files and png images of final records for all successfully detected clusters.

To work with your figure put figure.PNG to images/.

Regarding the TFOD API part, thanks to https://github.com/TannerGilbert/Tensorflow-Object-Detection-API-Train-Model

Citation

If you find this tool or any of it's parts useful, please cite our paper https://www.nature.com/articles/s41597-022-01699-3 You can also find more details regarding the usage of the algorithm there.