conditional_segmentation_thesis/scratchpad.py at main · bottyBotz/conditional_segmentation_thesis · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
import os, time
from config.global_train_config import config
import torch
import torch.optim as optim
import numpy as np
import random


from src.model.archs.condiSeg import condiSeg


model = condiSeg(config)


#model.get_input()
optimizer = optim.Adam(model.net.parameters(), lr=model.config.lr, weight_decay=1e-6)
for epoch in range(1, model.config.num_epochs + 1):
    for model.step, input_dict in enumerate(model.train_loader):
        #print(f"input_dict: {input_dict}")
        # for key in input_dict.keys():
        #     print(f"key: {key}")
        # for key, value in input_dict.items():
        #     print(f"{key}: {value}")

        fx_img, mv_img = input_dict['fx_img'], input_dict['mv_img']  # [batch, 1, x, y, z]  #used to be .cuda() after each input dict selection
        fx_seg, mv_seg = input_dict['fx_seg'], input_dict['mv_seg']  # label #used to be .cuda()
        print(">>> SHAPES BEFORE PROCESSING:\n")
        print(f"fx_img shape: {fx_img.shape}")
        print(f"mv_img shape: {mv_img.shape}")
        print(f"fx_seg shape: {fx_seg.shape}")
        print(f"mv_seg shape: {mv_seg.shape}")
        print('mv_seg.shape[0]: ', mv_seg.shape[0])
        print('mv_seg.shape[0] - 1: ', mv_seg.shape[0] - 1)
        print(">>> SHAPES AFTER TWO STAGE SAMPLING:\n")
        random_label_index = random.randint(0, mv_seg.shape[0]-1) #pick a random example in this batch
        moving_label, fixed_label = mv_seg[random_label_index], fx_seg[random_label_index] #
        print("random_label_index", random_label_index)
        print(f"mv_seg[random_label_index] (moving_label shape): {mv_seg[random_label_index].shape}")
        print(f"fx_seg[random_label_index] (fixed_label shape): {fx_seg[random_label_index].shape}")

        # print(">>> SHAPE WITH TWO STAGE SAMPLING:\n")
        # print(f"mv_seg[random_label_index]: {mv_seg[random_label_index].shape}")
        # print(f"torch.FloatTensor(moving_image[None, ...]).shape: {torch.FloatTensor(mv_img[None, ...]).shape}")
        # #print(f"torch.FloatTensor(moving_image[None, ...]: {torch.FloatTensor(mv_img[None, ...])})")

        print(">>> USING GET INPUT:\n")
        fx_img, fx_seg, mv_img, mv_seg = model.get_input(input_dict)
        print(">>> SHAPES BEFORE PROCESSING:\n")
        print(f"fx_img shape: {fx_img.shape}")
        print(f"mv_img shape: {mv_img.shape}")
        print(f"fx_seg shape: {fx_seg.shape}")
        print(f"mv_seg shape: {mv_seg.shape}")
        print('mv_seg.shape[0]: ', mv_seg.shape[0])
        print('mv_seg.shape[0] - 1: ', mv_seg.shape[0] - 1)
        print(">>> SHAPES AFTER TWO STAGE SAMPLING:\n")
        random_label_index = random.randint(0, mv_seg.shape[0]-1) #pick a random example in this batch
        moving_label, fixed_label = mv_seg[random_label_index], fx_seg[random_label_index] #
        print("random_label_index", random_label_index)
        print(f"mv_seg[random_label_index] (moving_label shape): {mv_seg[random_label_index].shape}")
        print(f"fx_seg[random_label_index] (fixed_label shape): {fx_seg[random_label_index].shape}")


        break
        model.net.zero_grad()
        output = model.net(input_dict)
        loss = model.loss(output, input_dict)
        loss.backward()
        optimizer.step()
        if model.step % model.config.print_freq == 0:
            print(f'Epoch: {epoch}, Step: {model.step}, Loss: {loss.item():.4f}')
    break

#python3 scratchpad.py --project ConditionalSeg --exp_name test_exp  --model CondiSegUNet --cv 0 --patient_cohort intra --crop_on_seg_aug 0 --use_pseudo_label 0 --w_dce 1.0 --w_bce 1 --input_shape 64 101 91 --data_path ./Data/fullResCropIntensityClip_resampled --batch_size 8 --using_HPC 0 --two_stage_sampling 0
#Data path on weiss
#../../../raid/candi/daniel/Data/others/deepRegData/fullResCropIntensityClip_resampled
#python3 scratchpad.py --project ConditionalSeg --exp_name test_exp  --model CondiSegUNet --cv 0 --patient_cohort intra --crop_on_seg_aug 0 --use_pseudo_label 0 --w_dce 1.0 --w_bce 1 --input_shape 64 101 91 --data_path ../../../raid/candi/daniel/Data/others/deepRegData/fullResCropIntensityClip_resampled --batch_size 8 --using_HPC 0 --two_stage_sampling 0

# scratchpad.py
# --project ConditionalSeg
# --exp_name test_exp
# --model CondiSegUNet
# --cv 0
# --patient_cohort intra
# --crop_on_seg_aug 0
# --use_pseudo_label 0
# --w_dce 1.0
# --w_bce 1
# --input_shape 64 101 91
# --data_path ../../../raid/candi/daniel/Data/others/deepRegData/fullResCropIntensityClip_resampled
# --batch_size 8
# --using_HPC 0
# --two_stage_sampling 0