我目前正在微调ssd mobilenet v2模型，以改善人工检测。
我的ssd_mobilenet_v2_coco_config代码为：
# SSD with Mobilenet v2 configuration for MSCOCO Dataset.
# Users should configure the fine_tune_checkpoint field in the train config as
# well as the label_map_path and input_path fields in the train_input_reader and
# eval_input_reader. Search for "PATH_TO_BE_CONFIGURED" to find the fields that
# should be configured.
model {
ssd {
num_classes: 1
box_coder {
faster_rcnn_box_coder {
y_scale: 10.0
x_scale: 10.0
height_scale: 5.0
width_scale: 5.0
}
}
matcher {
argmax_matcher {
matched_threshold: 0.5
unmatched_threshold: 0.5
ignore_thresholds: false
negatives_lower_than_unmatched: true
force_match_for_each_row: true
}
}
similarity_calculator {
iou_similarity {
}
}
anchor_generator {
ssd_anchor_generator {
num_layers: 6
min_scale: 0.2
max_scale: 0.95
aspect_ratios: 1.0
aspect_ratios: 2.0
aspect_ratios: 0.5
aspect_ratios: 3.0
aspect_ratios: 0.3333
}
}
image_resizer {
fixed_shape_resizer {
height: 300
width: 300
}
}
box_predictor {
convolutional_box_predictor {
min_depth: 0
max_depth: 0
num_layers_before_predictor: 0
use_dropout: false
dropout_keep_probability: 0.8
kernel_size: 3
box_code_size: 4
apply_sigmoid_to_scores: false
conv_hyperparams {
activation: RELU_6,
regularizer {
l2_regularizer {
weight: 0.00004
}
}
initializer {
truncated_normal_initializer {
stddev: 0.03
mean: 0.0
}
}
batch_norm {
train: true,
scale: true,
center: true,
decay: 0.9997,
epsilon: 0.001,
}
}
}
}
feature_extractor {
type: 'ssd_mobilenet_v2'
min_depth: 16
depth_multiplier: 1.0
use_depthwise: true
conv_hyperparams {
activation: RELU_6,
regularizer {
l2_regularizer {
weight: 0.00004
}
}
initializer {
truncated_normal_initializer {
stddev: 0.03
mean: 0.0
}
}
batch_norm {
train: true,
scale: true,
center: true,
decay: 0.9997,
epsilon: 0.001,
}
}
}
loss {
classification_loss {
weighted_sigmoid {
}
}
localization_loss {
weighted_smooth_l1 {
}
}
hard_example_miner {
num_hard_examples: 3000
iou_threshold: 0.99
loss_type: CLASSIFICATION
max_negatives_per_positive: 3
min_negatives_per_image: 3
}
classification_weight: 1.0
localization_weight: 1.0
}
normalize_loss_by_num_matches: true
post_processing {
batch_non_max_suppression {
score_threshold: 1e-8
iou_threshold: 0.6
max_detections_per_class: 100
max_total_detections: 100
}
score_converter: SIGMOID
}
}
}
train_config: {
batch_size: 24
optimizer {
rms_prop_optimizer: {
learning_rate: {
exponential_decay_learning_rate {
initial_learning_rate: 0.004
decay_steps: 800720
decay_factor: 0.95
}
}
momentum_optimizer_value: 0.9
decay: 0.9
epsilon: 1.0
}
}
fine_tune_checkpoint: "D:/Databases/Coco/cctv/tf/ssd_mobilenet_v2_coco_2018_03_29/model.ckpt"
fine_tune_checkpoint_type: "detection"
# Note: The below line limits the training process to 200K steps, which we
# empirically found to be sufficient enough to train the pets dataset. This
# effectively bypasses the learning rate schedule (the learning rate will
# never decay). Remove the below line to train indefinitely.
num_steps: 200000
data_augmentation_options {
random_horizontal_flip {
}
}
data_augmentation_options {
ssd_random_crop {
}
}
}
train_input_reader: {
tf_record_input_reader {
input_path: "D:/Code/Image/cctvmodel/tfrecordfinalALL/train2.record"
}
label_map_path: "D:/Code/Image/cctvmodel/label_map.pbtxt"
}
eval_config: {
num_examples: 8000
# Note: The below line limits the evaluation process to 10 evaluations.
# Remove the below line to evaluate indefinitely.
max_evals: 10
}
eval_input_reader: {
tf_record_input_reader {
input_path: "D:/Code/Image/cctvmodel/tfrecordfinalALL/test2.record"
}
label_map_path: "D:/Code/Image/cctvmodel/label_map.pbtxt"
shuffle: false
num_readers: 1
}
已配置该配置文件以进行再培训。要运行配置文件，请使用以下代码：
(tensorflow) c:\models-master\research>python object_detection/legacy/train.py --logtostderr --train_dir=training/model2/ --pipeline_config_path=training/ssd_mobilenet_v2_2.config
对于微调，我指的是tutorial 1。它指出要对模型进行微调，需要执行以下操作：
1）创建一个对象检测训练pipeline.config文件：从/path/to/pretrainedModels/faster_rcnn_resnet50_lowproposals_coco_2017_11_08/pipeline.config修改一个，只需更改num_classes, fine_tune_checkpoint, num_steps, label_map_path and tf_record_input_reader/input_path.
2）为from_detection_checkpoint设置正确的值。如果要从预先训练的对象检测模型中进行微调，请将其设置为true；否则，将其设置为true。如果来自分类预训练模型，则将其设置为false。
3）使用以下命令进行训练：
From the tensorflow/models/research/ directory
python object_detection/train.py \
--logtostderr \
--train_dir=${PATH_TO_TRAIN_DIR} \
--pipeline_config_path=${PATH_TO_YOUR_PIPELINE_CONFIG}
我注意到许多教程与上面列出的教程非常相似。但是，该指令似乎没有充分举例说明最佳微调过程。如果不得不考虑如何在Keras中进行微调，那么它安静而灵活且精心设计。
例如，如何状态化需要冻结的层并指定应训练层的哪些子集。另外，由于型号不同，这是SSD移动网络v2的最佳条件。在大多数情况下，上层是冻结的，后五层是微调的。
也就是说，我们是否要添加一个辍学层等。但是，如何使用配置文件来实现这一点。
需要设置哪些其他参数来优化微调？
也就是说，还有另一个示例可以通过python笔记本tutorial 3进行微调。哪种方法最合适？