Skip to content

Water level measurement

Overviews

Using SAM and SOLOv2 to detect and measure water level

Please install ecos_core-1.1.0 or newest version and AdelaiDet-0.2.0 beforehand. 

pip install ecos_core-1.1.0.zip
pip install AdelaiDet-0.2.0.zip

Prepare your dataset

# your dataset structure should be like this
datasets/
    -annotations/
        - instances_train2017.json
        - instances_val2017.json
    -train2017/
        -*.jpg
    -val2017/
        -*.jpg

Usages

1. For training, run:

python train.py --opt-solo "<PATH_TO_OPT_FILE>"

Example:

python train.py --opt-solo ./opt.json
  • PATH_TO_OPT_FILE is path to file opt.json

2. For testing, run:

python predict.py --input-path "<PATH_TO_TEST_IMG>" --opt-sam "<PATH_TO_SAM_OPT_FILE>" --opt-solo "<PATH_TO_SOLOV2_OPT_FILE>" --output-path "<PATH_TO_OUTPUT_IMG>"

or

python predict.py --input-path ./../../../riwa_v2/test/images/ADE_train_00004363.jpg --opt-sam ./../../ecos_core/SAM/opt.json --opt-solo ./opt.json --output-path test.png
  • PATH_TO_TEST_IMG should be where the image file are
  • PATH_TO_OUTPUT_IMG where result will be saved
  • PATH_TO_SAM_OPT_FILE is path to file opt.json of SAM
  • PATH_TO_SOLOV2_OPT_FILE is path to file opt.json of SOLOv2
  • If the prediction is not good, please consider tuning the threshold or adding more data