def local_eval(model, test_loader=None, test_label_file=None):
prediction_file = 'pred_train.txt'
feed_infer(
prediction_file, lambda root_path: _infer(
model, root_path, test_loader=test_loader, local_val=True))
metric_result = evaluation_metrics(prediction_file, test_label_file)
print('Eval result: {:.4f}'.format(metric_result))
return metric_result
def local_eval(model, data_loader=None):
prediction_file = 'test_pred'
feed_infer(
prediction_file,
lambda root_path: _infer(model, root_path, data_loader=data_loader))
test_label_file = '/home/data/nipa_faces_sr_tmp2/test/test_label' # local datapath
metric_result = evaluation_metrics(prediction_file, test_label_file)
print('Eval result: {:.4f}'.format(metric_result))
return metric_result
def local_eval(model, test_loader=None, test_label_file=None):
model.eval()
prediction_file = 'pred_train.txt'
feed_infer(prediction_file, lambda root_path: _infer(model, root_path, test_loader=test_loader))
if not test_label_file:
test_label_file = os.path.join(VAL_DATASET_PATH, 'test_label')
metric_result = evaluation_metrics(
prediction_file,
test_label_file)
return metric_result
def local_eval(model, test_loader=None, test_label_file=None):
prediction_file = 'pred_train.txt'
feed_infer(
prediction_file,
lambda root_path: _infer(model, root_path, test_loader=test_loader))
if not test_label_file:
test_label_file = os.path.join(VAL_DATASET_PATH, 'test_label')
metric_result = evaluation_metrics(prediction_file, test_label_file)
logger.info('Eval result: {:.4f} mIoU'.format(metric_result))
return metric_result
def local_eval(model,
test_loader=None,
test_label_file="../food_img/food_label.txt"):
prediction_file = "pred_train.txt"
feed_infer(
prediction_file,
lambda root_path: _infer(
model, root_path, test_loader=test_loader, local_val=True),
)
metric_result = evaluation_metrics(prediction_file, test_label_file)
print("Eval result: {:.4f}".format(metric_result))
return metric_result
def local_eval(model, test_loader=None, test_label_file=None):
"""Local debugging function.
You can use this function for debugging.
"""
prediction_file = 'pred_train.txt'
feed_infer(
prediction_file,
lambda root_path: _infer(model, root_path, test_loader=test_loader))
if not test_label_file:
test_label_file = os.path.join(VAL_DATASET_PATH, 'test_label')
metric_result = evaluation_metrics(prediction_file, test_label_file)
print('[Eval result] recall: {:.2f}'.format(metric_result))
return metric_result
def local_eval(model, loader, gt_path):
"""Local debugging function.
You can use this function for debugging. You may need dummy gt file.
Args:
model: instance.
test_loader: instance.
gt_path: string.
Returns:
metric_result: float. Performance of your method.
"""
pred_path = 'pred.txt'
feed_infer(
pred_path, lambda root_path: _infer(
model=model, root_path=root_path, loader=loader))
metric_result = evaluation_metrics(pred_path, gt_path)
return metric_result
