获取人脸数据集

实验六/get_dataset.py

@@ -16,5 +16,5 @@ def save_img_subset(data, save_path, num_samples):
         img.save(os.path.join(save_path, f"{i}-label-{label}.png"))
 
 # 保存前 600 张训练集图片和前 100 张测试集图片
-save_img_subset(train_data, './DataImages-Train', 6000)
-save_img_subset(test_data, './DataImages-Test', 1000)
+save_img_subset(train_data, './DataImages-Train', 60000)
+save_img_subset(test_data, './DataImages-Test', 10000)

实验六/train.py

@@ -64,8 +64,8 @@ def ensure_dir_exists(directory):
         os.makedirs(directory)
 
 # 加载训练数据
-trains_paths, trains_labels = load_data("cache/pretrains/train")
-test_paths, test_labels = load_data("cache/pretrains/test")
+trains_paths, trains_labels = load_data("DataImages-Train")
+test_paths, test_labels = load_data("DataImages-Test")
 
 # 提取特征和标签
 X_train = np.array([extract_features(train_path) for train_path in tqdm.tqdm(trains_paths, desc="训练集特征提取中：")])
@@ -87,7 +87,7 @@ for test_sample in tqdm.tqdm(X_test, desc="测试集中预测进度"):
     Y_pred.append(classifier.predict(test_sample.reshape(1, -1)))
 accuracy = accuracy_score(Y_test, Y_pred)
 
-print(f"性能: {accuracy * 100:.2f}%")
+print(f"准确率: {accuracy * 100:.2f}%")
 
 # 保存模型
 ensure_dir_exists("models")