作者: Sayak Paul
创建日期 2021/05/28
最后修改日期 2023/12/08
描述:使用迁移学习和循环模型在 UCF101 数据集上训练视频分类器。
此示例演示了视频分类,这是一个重要的用例,在推荐、安全等领域有应用。我们将使用 UCF101 数据集 来构建我们的视频分类器。该数据集包含分类到不同动作的视频,例如板球击球、拳击、骑自行车等等。该数据集通常用于构建动作识别器,它是视频分类的一种应用。
视频由一组有序的帧组成。每个帧包含空间信息,而这些帧的序列包含时间信息。为了对这两个方面进行建模,我们使用了一个混合架构,该架构包含卷积(用于空间处理)以及循环层(用于时间处理)。具体来说,我们将使用卷积神经网络 (CNN) 和循环神经网络 (RNN),RNN 由 GRU 层 组成。这种混合架构通常被称为 **CNN-RNN**。
此示例需要 TensorFlow 2.5 或更高版本,以及 TensorFlow 文档,可以使用以下命令安装
!pip install -q git+https://github.com/tensorflow/docs
为了使此示例的运行时间相对较短,我们将使用原始 UCF101 数据集的子采样版本。你可以参考 这个笔记本 以了解如何进行子采样。
!!wget -q https://github.com/sayakpaul/Action-Recognition-in-TensorFlow/releases/download/v1.0.0/ucf101_top5.tar.gz
!tar xf ucf101_top5.tar.gz
import os
import keras
from imutils import paths
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import imageio
import cv2
from IPython.display import Image
IMG_SIZE = 224
BATCH_SIZE = 64
EPOCHS = 10
MAX_SEQ_LENGTH = 20
NUM_FEATURES = 2048
train_df = pd.read_csv("train.csv")
test_df = pd.read_csv("test.csv")
print(f"Total videos for training: {len(train_df)}")
print(f"Total videos for testing: {len(test_df)}")
train_df.sample(10)
Total videos for training: 594
Total videos for testing: 224
video_name | tag | |
---|---|---|
492 | v_TennisSwing_g10_c03.avi | TennisSwing |
536 | v_TennisSwing_g16_c05.avi | TennisSwing |
413 | v_ShavingBeard_g16_c05.avi | ShavingBeard |
268 | v_Punch_g12_c04.avi | Punch |
288 | v_Punch_g15_c03.avi | Punch |
30 | v_CricketShot_g12_c03.avi | CricketShot |
449 | v_ShavingBeard_g21_c07.avi | ShavingBeard |
524 | v_TennisSwing_g14_c07.avi | TennisSwing |
145 | v_PlayingCello_g12_c01.avi | PlayingCello |
566 | v_TennisSwing_g21_c03.avi | TennisSwing |
训练视频分类器的一个挑战是如何将视频馈送到网络。 这篇博文 讨论了五种这样的方法。由于视频是按顺序排列的帧,我们可以提取帧并将它们放入一个 3D 张量中。但是视频的帧数可能不同,这会阻止我们将其堆叠成批次(除非我们使用填充)。作为替代方案,我们可以 **以固定间隔保存视频帧,直到达到最大帧数为止**。在本示例中,我们将执行以下操作
请注意,此工作流程与 涉及文本序列的问题 相同。UCF101 数据集中的视频 已知 在帧之间没有极端的对象和动作变化。因此,对于学习任务而言,仅考虑少数帧可能就可以了。但是这种方法可能无法很好地泛化到其他视频分类问题。我们将使用 OpenCV 的 VideoCapture()
方法 从视频中读取帧。
# The following two methods are taken from this tutorial:
# https://tensorflowcn.cn/hub/tutorials/action_recognition_with_tf_hub
def crop_center_square(frame):
y, x = frame.shape[0:2]
min_dim = min(y, x)
start_x = (x // 2) - (min_dim // 2)
start_y = (y // 2) - (min_dim // 2)
return frame[start_y : start_y + min_dim, start_x : start_x + min_dim]
def load_video(path, max_frames=0, resize=(IMG_SIZE, IMG_SIZE)):
cap = cv2.VideoCapture(path)
frames = []
try:
while True:
ret, frame = cap.read()
if not ret:
break
frame = crop_center_square(frame)
frame = cv2.resize(frame, resize)
frame = frame[:, :, [2, 1, 0]]
frames.append(frame)
if len(frames) == max_frames:
break
finally:
cap.release()
return np.array(frames)
我们可以使用预训练的网络从提取的帧中提取有意义的特征。 Keras Applications
模块提供了许多在 ImageNet-1k 数据集 上预训练的最先进模型。我们将使用 InceptionV3 模型 来实现这一目的。
def build_feature_extractor():
feature_extractor = keras.applications.InceptionV3(
weights="imagenet",
include_top=False,
pooling="avg",
input_shape=(IMG_SIZE, IMG_SIZE, 3),
)
preprocess_input = keras.applications.inception_v3.preprocess_input
inputs = keras.Input((IMG_SIZE, IMG_SIZE, 3))
preprocessed = preprocess_input(inputs)
outputs = feature_extractor(preprocessed)
return keras.Model(inputs, outputs, name="feature_extractor")
feature_extractor = build_feature_extractor()
视频的标签是字符串。神经网络不理解字符串值,因此在将它们馈送到模型之前必须将其转换为某种数值形式。这里我们将使用 StringLookup
层将类标签编码为整数。
label_processor = keras.layers.StringLookup(
num_oov_indices=0, vocabulary=np.unique(train_df["tag"])
)
print(label_processor.get_vocabulary())
['CricketShot', 'PlayingCello', 'Punch', 'ShavingBeard', 'TennisSwing']
最后,我们可以将所有部分组合在一起以创建我们的数据处理工具。
def prepare_all_videos(df, root_dir):
num_samples = len(df)
video_paths = df["video_name"].values.tolist()
labels = df["tag"].values
labels = keras.ops.convert_to_numpy(label_processor(labels[..., None]))
# `frame_masks` and `frame_features` are what we will feed to our sequence model.
# `frame_masks` will contain a bunch of booleans denoting if a timestep is
# masked with padding or not.
frame_masks = np.zeros(shape=(num_samples, MAX_SEQ_LENGTH), dtype="bool")
frame_features = np.zeros(
shape=(num_samples, MAX_SEQ_LENGTH, NUM_FEATURES), dtype="float32"
)
# For each video.
for idx, path in enumerate(video_paths):
# Gather all its frames and add a batch dimension.
frames = load_video(os.path.join(root_dir, path))
frames = frames[None, ...]
# Initialize placeholders to store the masks and features of the current video.
temp_frame_mask = np.zeros(
shape=(
1,
MAX_SEQ_LENGTH,
),
dtype="bool",
)
temp_frame_features = np.zeros(
shape=(1, MAX_SEQ_LENGTH, NUM_FEATURES), dtype="float32"
)
# Extract features from the frames of the current video.
for i, batch in enumerate(frames):
video_length = batch.shape[0]
length = min(MAX_SEQ_LENGTH, video_length)
for j in range(length):
temp_frame_features[i, j, :] = feature_extractor.predict(
batch[None, j, :], verbose=0,
)
temp_frame_mask[i, :length] = 1 # 1 = not masked, 0 = masked
frame_features[idx,] = temp_frame_features.squeeze()
frame_masks[idx,] = temp_frame_mask.squeeze()
return (frame_features, frame_masks), labels
train_data, train_labels = prepare_all_videos(train_df, "train")
test_data, test_labels = prepare_all_videos(test_df, "test")
print(f"Frame features in train set: {train_data[0].shape}")
print(f"Frame masks in train set: {train_data[1].shape}")
Frame features in train set: (594, 20, 2048)
Frame masks in train set: (594, 20)
上面的代码块将需要大约 20 分钟才能执行,具体取决于执行它的机器。
现在,我们可以将这些数据馈送到由循环层(如 GRU
)组成的序列模型。
# Utility for our sequence model.
def get_sequence_model():
class_vocab = label_processor.get_vocabulary()
frame_features_input = keras.Input((MAX_SEQ_LENGTH, NUM_FEATURES))
mask_input = keras.Input((MAX_SEQ_LENGTH,), dtype="bool")
# Refer to the following tutorial to understand the significance of using `mask`:
# https://keras.org.cn/api/layers/recurrent_layers/gru/
x = keras.layers.GRU(16, return_sequences=True)(
frame_features_input, mask=mask_input
)
x = keras.layers.GRU(8)(x)
x = keras.layers.Dropout(0.4)(x)
x = keras.layers.Dense(8, activation="relu")(x)
output = keras.layers.Dense(len(class_vocab), activation="softmax")(x)
rnn_model = keras.Model([frame_features_input, mask_input], output)
rnn_model.compile(
loss="sparse_categorical_crossentropy", optimizer="adam", metrics=["accuracy"]
)
return rnn_model
# Utility for running experiments.
def run_experiment():
filepath = "/tmp/video_classifier/ckpt.weights.h5"
checkpoint = keras.callbacks.ModelCheckpoint(
filepath, save_weights_only=True, save_best_only=True, verbose=1
)
seq_model = get_sequence_model()
history = seq_model.fit(
[train_data[0], train_data[1]],
train_labels,
validation_split=0.3,
epochs=EPOCHS,
callbacks=[checkpoint],
)
seq_model.load_weights(filepath)
_, accuracy = seq_model.evaluate([test_data[0], test_data[1]], test_labels)
print(f"Test accuracy: {round(accuracy * 100, 2)}%")
return history, seq_model
_, sequence_model = run_experiment()
Epoch 1/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.3058 - loss: 1.5597
Epoch 1: val_loss improved from inf to 1.78077, saving model to /tmp/video_classifier/ckpt.weights.h5
13/13 ━━━━━━━━━━━━━━━━━━━━ 2s 36ms/step - accuracy: 0.3127 - loss: 1.5531 - val_accuracy: 0.1397 - val_loss: 1.7808
Epoch 2/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.5216 - loss: 1.2704
Epoch 2: val_loss improved from 1.78077 to 1.78026, saving model to /tmp/video_classifier/ckpt.weights.h5
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 13ms/step - accuracy: 0.5226 - loss: 1.2684 - val_accuracy: 0.1788 - val_loss: 1.7803
Epoch 3/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.6189 - loss: 1.1656
Epoch 3: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.6174 - loss: 1.1651 - val_accuracy: 0.2849 - val_loss: 1.8322
Epoch 4/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.6518 - loss: 1.0645
Epoch 4: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 13ms/step - accuracy: 0.6515 - loss: 1.0647 - val_accuracy: 0.2793 - val_loss: 2.0419
Epoch 5/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.6833 - loss: 0.9976
Epoch 5: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.6843 - loss: 0.9965 - val_accuracy: 0.3073 - val_loss: 1.9077
Epoch 6/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.7229 - loss: 0.9312
Epoch 6: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.7241 - loss: 0.9305 - val_accuracy: 0.3017 - val_loss: 2.1513
Epoch 7/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.8023 - loss: 0.9132
Epoch 7: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.8035 - loss: 0.9093 - val_accuracy: 0.3184 - val_loss: 2.1705
Epoch 8/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.8127 - loss: 0.8380
Epoch 8: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.8128 - loss: 0.8356 - val_accuracy: 0.3296 - val_loss: 2.2043
Epoch 9/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.8494 - loss: 0.7641
Epoch 9: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.8494 - loss: 0.7622 - val_accuracy: 0.3017 - val_loss: 2.3734
Epoch 10/10
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 9ms/step - accuracy: 0.8634 - loss: 0.6883
Epoch 10: val_loss did not improve from 1.78026
13/13 ━━━━━━━━━━━━━━━━━━━━ 0s 12ms/step - accuracy: 0.8649 - loss: 0.6882 - val_accuracy: 0.3240 - val_loss: 2.4410
7/7 ━━━━━━━━━━━━━━━━━━━━ 0s 3ms/step - accuracy: 0.7816 - loss: 1.0624
Test accuracy: 56.7%
注意:为了使此示例的运行时间相对较短,我们只使用了一些训练示例。这个训练示例数量相对于所用序列模型来说很少,该模型具有 99,909 个可训练参数。鼓励您从 UCF101 数据集中使用 上面提到的笔记本 采样更多数据,并训练同一个模型。
def prepare_single_video(frames):
frames = frames[None, ...]
frame_mask = np.zeros(
shape=(
1,
MAX_SEQ_LENGTH,
),
dtype="bool",
)
frame_features = np.zeros(shape=(1, MAX_SEQ_LENGTH, NUM_FEATURES), dtype="float32")
for i, batch in enumerate(frames):
video_length = batch.shape[0]
length = min(MAX_SEQ_LENGTH, video_length)
for j in range(length):
frame_features[i, j, :] = feature_extractor.predict(batch[None, j, :])
frame_mask[i, :length] = 1 # 1 = not masked, 0 = masked
return frame_features, frame_mask
def sequence_prediction(path):
class_vocab = label_processor.get_vocabulary()
frames = load_video(os.path.join("test", path))
frame_features, frame_mask = prepare_single_video(frames)
probabilities = sequence_model.predict([frame_features, frame_mask])[0]
for i in np.argsort(probabilities)[::-1]:
print(f" {class_vocab[i]}: {probabilities[i] * 100:5.2f}%")
return frames
# This utility is for visualization.
# Referenced from:
# https://tensorflowcn.cn/hub/tutorials/action_recognition_with_tf_hub
def to_gif(images):
converted_images = images.astype(np.uint8)
imageio.mimsave("animation.gif", converted_images, duration=100)
return Image("animation.gif")
test_video = np.random.choice(test_df["video_name"].values.tolist())
print(f"Test video path: {test_video}")
test_frames = sequence_prediction(test_video)
to_gif(test_frames[:MAX_SEQ_LENGTH])
Test video path: v_TennisSwing_g03_c01.avi
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 35ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 35ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 34ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 32ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 33ms/step
1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 166ms/step
CricketShot: 46.99%
ShavingBeard: 18.83%
TennisSwing: 14.65%
Punch: 12.41%
PlayingCello: 7.12%
<IPython.core.display.Image object>
keras.applications
中的其他模型。MAX_SEQ_LENGTH
的不同组合,观察这将如何影响性能。