python - 在统计上成功的训练之后,结果并没有通过实时摄像机反映出来。我该如何解决这个问题?
问题描述
我一直在使用 TensorFlow 在 Linux 上进行牛奶盒检测。使用的语言是python。正在使用的模型是 faster_rcnn_inception_v2_pets。我是机器学习的新手。这是我第一次学习如何进行对象检测,请帮助!
我们的数据集:我们在相同的环境(冰箱架)中分别拍摄了每个牛奶盒的照片(每个牛奶盒大约 130 张照片)。我们拍摄了多个牛奶盒的混合照片(大约 400 张照片)。
以下是我们给纸箱贴标签的方法:使用 OpenLabeler
这是不准确的实时摄像头馈送(问题):证据
编码:
# Faster R-CNN with Inception v2, configured for Oxford-IIIT Pets 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 {
faster_rcnn {
num_classes: 6
image_resizer {
keep_aspect_ratio_resizer {
min_dimension: 600
max_dimension: 1024
}
}
feature_extractor {
type: 'faster_rcnn_inception_v2'
first_stage_features_stride: 16
}
first_stage_anchor_generator {
grid_anchor_generator {
scales: [0.25, 0.5, 1.0, 2.0]
aspect_ratios: [0.5, 1.0, 2.0]
height_stride: 16
width_stride: 16
}
}
first_stage_box_predictor_conv_hyperparams {
op: CONV
regularizer {
l2_regularizer {
weight: 0.0
}
}
initializer {
truncated_normal_initializer {
stddev: 0.01
}
}
}
first_stage_nms_score_threshold: 0.0
first_stage_nms_iou_threshold: 0.7
first_stage_max_proposals: 300
first_stage_localization_loss_weight: 2.0
first_stage_objectness_loss_weight: 1.0
initial_crop_size: 14
maxpool_kernel_size: 2
maxpool_stride: 2
second_stage_box_predictor {
mask_rcnn_box_predictor {
use_dropout: false
dropout_keep_probability: 1.0
fc_hyperparams {
op: FC
regularizer {
l2_regularizer {
weight: 0.0
}
}
initializer {
variance_scaling_initializer {
factor: 1.0
uniform: true
mode: FAN_AVG
}
}
}
}
}
second_stage_post_processing {
batch_non_max_suppression {
score_threshold: 0.0
iou_threshold: 0.6
max_detections_per_class: 100
max_total_detections: 300
}
score_converter: SOFTMAX
}
second_stage_localization_loss_weight: 2.0
second_stage_classification_loss_weight: 1.0
}
}
train_config: {
batch_size: 1
optimizer {
momentum_optimizer: {
learning_rate: {
exponential_decay_learning_rate {
initial_learning_rate: 0.0002
decay_steps: 5000
decay_factor: 0.9
}
}
momentum_optimizer_value: 0.9
}
use_moving_average: false
}
gradient_clipping_by_norm: 10.0
fine_tune_checkpoint: "/home/konbini/tensorflow1/models/research/object_detection/faster_rcnn_inception_v2_coco_2018_01_28/model.ckpt"
from_detection_checkpoint: true
# 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: 150000
data_augmentation_options {
random_horizontal_flip {
}
}
}
train_input_reader: {
tf_record_input_reader {
input_path: "/home/tensorflow1/models/research/object_detection/train.record"
}
label_map_path: "/home/tensorflow1/models/research/object_detection/training/labelmap.pbtxt"
}
eval_config: {
num_examples: 288
# Number of images in testing folder
# Note: The below line limits the evaluation process to 10 evaluations.
# Remove the below line to evaluate indefinitely.
}
eval_input_reader: {
tf_record_input_reader {
input_path: "/home/tensorflow1/models/research/object_detection/test.record"
}
label_map_path: "/home/tensorflow1/models/research/object_detection/training/labelmap.pbtxt"
shuffle: false
num_readers: 1
}
当我们对任何类型的颜色变化进行增强时,结果会变得更糟。当我们对任何类型的旋转进行增强时,结果都是一样的。证据
我有两个问题:我们的数据集有问题吗?我们的标注方法有问题吗?
解决方案
这可能是因为您的模型过度拟合了训练数据集,因此没有足够泛化来处理测试数据集。我可以看到您没有使用增强技术,除了random_horizontal_flip
在您的配置文件中可能有助于更好地概括您的模型。train_config
您可以在如下部分添加增强技术:
train_config: {
...
data_augmentation_options {
random_rotation90 {
}
}
data_augmentation_options {
random_distort_color {
}
}
data_augmentation_options {
random_adjust_brightness {
}
}
data_augmentation_options {
random_adjust_contrast {
}
}
.....
}
所有可用的增强技术都可以在preprocessor.proto中找到:
推荐阅读
- python - pygame 在条件下自动重复声音
- c++ - 保持函数参数的 consteval-ness
- python - 跨多个维度重复一维数组中的值
- javascript - React路由器dom更改url,不渲染组件
- python - 我无法在 Windows 11 上安装 pyaudio?
- flutter - Flutter body 可能正常完成,导致返回 'null',但返回类型是潜在的不可为空的类型
- python - 值错误:'{{node conv2d_19/Conv2D}} = Conv2D[T=DT_FLOAT, data_format="NHWC" 从 4 中减去 8 导致的负维度大小
- python - 按任务将熊猫数据框行拆分为多行
- c - K&R 存储分配器说明
- php - PHP - 管理 URL