Batch Prediction¶

1. Download demo data¶

cd DeepDenoiser
wget https://github.com/wayneweiqiang/PhaseNet/releases/download/test_data/test_data.zip
unzip test_data.zip

2. Run batch prediction¶

DeepDenoiser currently supports three data formats: numpy, and mseed

• For numpy format:

  python deepdenoiser/predict.py --model_dir=model/190614-104802 --data_list=test_data/npz.csv --data_dir=test_data/npz --format=numpy --save_signal --plot_figure
  

• For mseed format:

  python deepdenoiser/predict.py --model_dir=model/190614-104802 --data_list=test_data/mseed.csv --data_dir=test_data/mseed --format=mseed --save_signal --plot_figure
  

Optional arguments:

usage: predict.py [-h] [--format FORMAT]
                  [--batch_size BATCH_SIZE] [--output_dir OUTPUT_DIR]
                  [--model_dir MODEL_DIR] [--sampling_rate SAMPLING_RATE]
                  [--data_dir DATA_DIR] [--data_list DATA_LIST]
                  [--plot_figure] [--save_signal] [--save_noise]

optional arguments:
  -h, --help            show this help message and exit
  --format FORMAT       Input data format: numpy or mseed
  --batch_size BATCH_SIZE
                        Batch size
  --output_dir OUTPUT_DIR
                        Output directory (default: output)
  --model_dir MODEL_DIR
                        Checkpoint directory (default: None)
  --sampling_rate SAMPLING_RATE
                        sampling rate of pred data
  --data_dir DATA_DIR   Input file directory
  --data_list DATA_LIST
                        Input csv file
  --plot_figure         If plot figure
  --save_signal         If save denoised signal
  --save_noise          If save denoised noise

3. Read denoised signals¶

In [1]:

import numpy as np
import matplotlib.pyplot as plt
import os
import glob
PROJECT_ROOT = os.path.realpath(os.path.join(os.path.abspath(''), ".."))

import numpy as np import matplotlib.pyplot as plt import os import glob PROJECT_ROOT = os.path.realpath(os.path.join(os.path.abspath(''), ".."))

In [2]:

plt.close("all")
for i, fp in enumerate(sorted(glob.glob(os.path.join(PROJECT_ROOT, "output/results/*npz")))):
    signal = np.load(fp)["data"][5000:8000,-1,-1]
    raw_signal = np.load(os.path.join(PROJECT_ROOT, "test_data/npz/", fp.split("/")[-1]))["data"][5000:8000,-1]
    plt.figure(figsize=(6,2))
    plt.subplot(121)
    plt.plot(raw_signal, 'k', linewidth=0.5)
    ylim = plt.ylim()
    plt.subplot(122)
    plt.plot(signal, 'k', linewidth=0.5)
    plt.ylim(ylim)
    plt.suptitle(fp.split("/")[-1])
    plt.tight_layout()
    plt.show()
    if i >= 3:
        break

plt.close("all") for i, fp in enumerate(sorted(glob.glob(os.path.join(PROJECT_ROOT, "output/results/*npz")))): signal = np.load(fp)["data"][5000:8000,-1,-1] raw_signal = np.load(os.path.join(PROJECT_ROOT, "test_data/npz/", fp.split("/")[-1]))["data"][5000:8000,-1] plt.figure(figsize=(6,2)) plt.subplot(121) plt.plot(raw_signal, 'k', linewidth=0.5) ylim = plt.ylim() plt.subplot(122) plt.plot(signal, 'k', linewidth=0.5) plt.ylim(ylim) plt.suptitle(fp.split("/")[-1]) plt.tight_layout() plt.show() if i >= 3: break

In [ ]: