apply upsampling before the network

.gitignore

@@ -9,4 +9,6 @@ preprocessed/
 picked_fb_test_data/
 example/checkpoints/
 
-*.egg-info
+*.egg-info
+0_fbp_env/
+build/

example/0_ex_train.py

@@ -1,14 +1,13 @@
 #%% ========== Loading required packages 
 import pandas as pd
-import sys
 import os.path as osp
 import matplotlib.pyplot as plt
 from time import time
 from pathlib import Path
 import shutil
 from matplotlib.ticker import MaxNLocator
-
-sys.path.append(osp.abspath(osp.join(__file__, "../../")))
+# import sys
+# sys.path.append(osp.abspath(osp.join(__file__, "../../")))
 from first_break_picking import train
 from first_break_picking.tools import seed_everything
 from first_break_picking.data import save_shots_fb

example/1_ex_predict.py

@@ -1,10 +1,10 @@
 #%% ========== Loading required packages 
-import sys
 import os.path as osp
 from pathlib import Path
 import shutil
 
-sys.path.append(osp.abspath(osp.join(__file__, "../../")))
+# import sys
+# sys.path.append(osp.abspath(osp.join(__file__, "../../")))
 from first_break_picking.data import save_shots_fb
 from first_break_picking import predict
 

example/config.py

@@ -1,5 +1,5 @@
 #%% ============ Initiate ===============
-num_epcohs = 5
+num_epcohs = 20
 
 batch_size = 15
 split_nt = 22

first_break_picking/train_eval/predict.py

@@ -347,7 +347,7 @@ def predict(base_dir: str,
     with torch.no_grad():
         if validation:
             true_masks = []
-            for shot_number, (batch, true_mask, fbt_file_name ) in enumerate(loop):
+            for shot_number, (batch, true_mask, fbt_file_name) in enumerate(loop):
                 fbt_file_name = fbt_file_name[0]
 
                 shot1, predicted_pick, predicted_segment, true_mask1 = predict_validation(
@@ -358,7 +358,7 @@ def predict(base_dir: str,
                     overlap=overlap,
                     shot_id=fbt_file_name,
                     smoothing_threshold=smoothing_threshold,
-                    upsampler=upsampler,
+                    # upsampler=upsampler,
                     data_info=data_info,
                     case_specific_parameters=case_specific_parameters
                 )
@@ -373,14 +373,18 @@ def predict(base_dir: str,
             for shot_number, (batch, fbt_file_name) in enumerate(loop):
                 fbt_file_name = fbt_file_name[0]
 
+                # nsp:  data.shape=[1, 3, 1, 512, 22]
+                batch, _ = upsampler(batch.squeeze(0), batch.squeeze(0))
+                #nsp:  data.shape= [3, 1, 512, 512])
+
                 shot, predicted_pick, predicted_segment = predict_test(
-                    batch=batch, 
+                    batch=batch.unsqueeze(0), 
                     model=model,
                     split_nt=split_nt,
                     overlap=overlap,
                     shot_id=fbt_file_name,
                     smoothing_threshold=smoothing_threshold,
-                    upsampler=upsampler,
+                    # upsampler=upsampler,
                     data_info=data_info,
                     case_specific_parameters=case_specific_parameters
                 )