def predict(self):
rights = []
labels = []
for inputs in self.test_action_data_set:
data, label = inputs
# data = data.unsqueeze(0) # 扩展一个维度
label = torch.LongTensor([int(label)])
if torch.cuda.is_available():
data = data.to(self.device)
labels.append(label)
output = self.model(data)
right = self.rightness(output, label)
rights.append(right)
# 计算校验集的平均准确度
right_ratio = 1.0 * np.sum([i[0] for i in rights]) / np.sum(
[i[1] for i in rights])
print("模式{}-{}-{},准确率：{:.3f},识别个数：{}".format(self.model_name,
self.axis,
self.batch_size,
right_ratio, len(labels)))
AUtils.metrics(np.array(labels),
np.array([i[3] for i in rights]).flatten())
AUtils.plot_confusion_matrix(
np.array(labels),
np.array([i[3] for i in rights]).flatten(),
classes=['Action0', 'Action1', 'Action2', 'Action3', 'Action4'],
savePath=
f'src/rec_batch/plt_img/{self.model_name}_{self.axis}_{self.batch_size}_predict.png',
title=f'{self.model_name}_{self.axis}_{self.batch_size}_predict')
def predict(self):
rights = []
labels = []
for data, label in self.test_action_data_set:
data = data.unsqueeze(0) # 扩展一个维度
label = torch.LongTensor([int(label)])
if torch.cuda.is_available():
data = data.cuda()
labels.append(label)
output = self.model(data)
right = self.rightness(output, label)
rights.append(right)
# 计算校验集的平均准确度
right_ratio = 1.0 * np.sum([i[0] for i in rights]) / np.sum(
[i[1] for i in rights])
print("模式{}-{},准确率：{:.3f},识别个数：{}".format(model_name, cls, right_ratio,
len(labels)))
AUtils.metrics(np.array(labels),
np.array([i[3] for i in rights]).flatten())
AUtils.plot_confusion_matrix(
np.array(labels),
np.array([i[3] for i in rights]).flatten(),
classes=[0, 1, 2, 3, 4],
savePath=
f'src/test_plt_img/{self.model_name}_{self.cls}_predict.png',
title=f'{self.model_name}_{self.cls}_predict')
def predict(self):
train_data, test_data = self.data
Xtest = test_data[:, :-1]
ytest = test_data[:, -1]
knn_model = joblib.load(fr'src/ml_cf_model/{self.model_name}_model-{self.axis}.pkl')
y_predict = knn_model.predict(Xtest)
AUtils.plot_confusion_matrix(ytest, y_predict, ['Action0', 'Action1', 'Action2', 'Action3', 'Action4'],
fr'src/ml_cf_plt_img/{self.model_name}_predict-{self.axis}.jpg',
title=fr'{self.model_name}-{self.axis} Confusion matrix')
AUtils.metrics(ytest, y_predict)
return ytest, y_predict