def main():
mode = "test" # test # debug
pos_loss_method = "mse" # "mse" # "smoothL1"
th_train_iou = 0.8 # 0.8
norm = True # False
min_test_epoch = 32
max_test_epoch = 32
loss_weight_lambda_ls = [0] # [0.1, 0.2, 0.5, 1, 2, 5, 10]
prevent_overfitting_method = "L2 Regu" # "L2 Regu" # "Dropout"
partial_l2 = False
dx_compute_method = "centre"
th_score = 0.8
test_arguments = TestArguments(norm, pos_loss_method, th_train_iou, min_test_epoch, max_test_epoch,
dx_compute_method=dx_compute_method,
score_threshold=th_score, partial_l2_penalty=partial_l2,
prevent_overfitting_method=prevent_overfitting_method, mode=mode)
test_arguments.show_arguments()
write_result_path = test_arguments.get_write_result_path()
all_csv_result = open(write_result_path, "w")
first_write(all_csv_result)
th_nms_iou_ls = [0.01]
th_iou_p_ls = [1]
for loss_weight_lambda in loss_weight_lambda_ls:
for th_iou_p in th_iou_p_ls:
for th_nms_iou in th_nms_iou_ls:
test_arguments.loss_weight_lambda = loss_weight_lambda
test_arguments.th_iou_p = th_iou_p
test_arguments.th_nms_iou = th_nms_iou
_test(test_arguments, all_csv_result)
def main():
pos_loss_method = "mse" # "mse"
th_train_iou = 0.8 # 0.8
norm = True # False
min_test_epoch = 31
max_test_epoch = 31
loss_weight_lambdas = [0] # [0.1, 0.2, 0.5, 1, 2, 5, 10]
prevent_overfitting_method = "Dropout" # "L2 Regu" # "Dropout"
partial_l2 = False
dx_compute_method = "centre"
output_flow = False
th_score = 0.8
for loss_weight_lambda in loss_weight_lambdas:
test_arguments = TestAruguments(norm, pos_loss_method, th_train_iou, min_test_epoch, max_test_epoch,
dx_compute_method=dx_compute_method, output_flow=output_flow,
score_threshold=th_score,
loss_weight_lambda=loss_weight_lambda, partial_l2_penalty=partial_l2,
prevent_overfitting_method=prevent_overfitting_method)
test_arguments.show_arguments()
write_result_path = test_arguments.get_write_result_path()
all_csv_result = open(write_result_path, "w")
first_write(all_csv_result)
th_nms_iou_ls = [0.01]
th_iou_p_ls = [1] # [0.6, 0.8, 1]
for th_iou_p in th_iou_p_ls:
for th_nms_iou in th_nms_iou_ls:
file_before = open("Debug Before NMS.txt", "w")
file_after = open("Debug After NMS.txt", "w")
test_arguments.th_nms_iou = th_nms_iou
test_arguments.th_iou_p = th_iou_p
_test_k_fold(file_before, file_after, test_arguments, all_csv_result)
def main():
pos_loss_method = "mse" # "mse"
th_train_iou = 0.6 # 0.8
norm = True # Flase
min_test_epoch = 40
max_test_epoch = 40
loss_weight_lambda = 1.0
prevent_overfitting_method = "Dropout" # "L2 Regu" # "Dropout"
partial_l2 = False
dx_compute_method = "centre" # "left_boundary" # "centre"
test_arguments = TestAruguments(
norm,
pos_loss_method,
th_train_iou,
min_test_epoch,
max_test_epoch,
dx_compute_method=dx_compute_method,
loss_weight_lambda=loss_weight_lambda,
partial_l2_penalty=partial_l2,
prevent_overfitting_method=prevent_overfitting_method)
test_arguments.show_arguments()
write_result_path = test_arguments.get_write_result_path()
all_csv_result = open(write_result_path, "w")
first_write(all_csv_result)
th_nms_iou_ls = [0.01]
th_iou_p_ls = [0.6, 0.8, 1] # [0.8] # [0.6, 0.8, 1] # [1],
for th_iou_p in th_iou_p_ls:
for th_nms_iou in th_nms_iou_ls:
test_arguments.th_nms_iou = th_nms_iou
test_arguments.th_iou_p = th_iou_p
_test_k_fold(test_arguments, all_csv_result)
def main():
th_train_iou = 1
write_file_path = "./result/insurance/normed_data/gt_train" + str(
th_train_iou) + ".csv"
all_csv_result = open(write_file_path, "w")
first_write(all_csv_result)
# th_iou_ls = [0.4, 0.5]
th_nms_iou_ls = [0.2]
# th_nms_iou_ls = [0.2]
th_iou_p_ls = [0.6, 0.8, 1] # [1] #
# for th_iou in th_iou_ls:
print("For this K fold test:\n")
print("th_train_iou: ", th_train_iou, "\n\n")
for th_iou_p in th_iou_p_ls:
for th_nms_iou in th_nms_iou_ls:
model_k_folder_path = "model/rcnn_jieba/relabeled_gt_model_" + str(
th_train_iou) + "/" # gt train
test_sentence_npz_k_folder_path = "dataset/test/test_relabeled_data_npz/"
_test_k_fold(model_k_folder_path, test_sentence_npz_k_folder_path,
th_nms_iou, th_iou_p, all_csv_result)
def main():
pos_loss_method = "mse" # "mse"
th_train_iou = 0.6 # 0.8
norm = True # Flase
min_test_epoch = 40
max_test_epoch = 40
loss_weight_lambda = 1.0
prevent_overfitting_method = "Dropout" # "L2 Regu" # "Dropout"
partial_l2 = False
dx_compute_method = "centre"
output_flow = True
th_score = 0.6
test_arguments = TestAruguments(
norm,
pos_loss_method,
th_train_iou,
min_test_epoch,
max_test_epoch,
dx_compute_method=dx_compute_method,
output_flow=output_flow,
score_threshold=th_score,
loss_weight_lambda=loss_weight_lambda,
partial_l2_penalty=partial_l2,
prevent_overfitting_method=prevent_overfitting_method)
test_arguments.show_arguments()
write_result_path = test_arguments.get_write_result_path()
all_csv_result = open(write_result_path, "w")
first_write(all_csv_result)
th_nms_iou_ls = [0.4]
th_iou_p_ls = [0.6, 0.8, 1]
for th_iou_p in th_iou_p_ls:
for th_nms_iou in th_nms_iou_ls:
file = open("Before NMS.txt", "w")
test_arguments.th_nms_iou = th_nms_iou
test_arguments.th_iou_p = th_iou_p
_test_k_fold(file, test_arguments, all_csv_result)