Я новичок. У меня есть проблема, которую я не могу решить, и я очень в беде. Я хочу, чтобы ты сказал мне. Я хочу обнаружить объект на моем Raspberry Pi. Я хочу выполнить модель вывода на Edge TPU.
1.Создать новую модель из изученной модели с тонкой настройкой.
2.Скомпилировать для работы с EdgeTPU и развернуть на RaspberryPi.
Я попробовал две вещи для достижения этой цели, но они не увенчались успехом.
Первая
1. Модель обнаружения тензора. Для тонкой настройки я использовал zs ssd_mobilenet_v3_large_coco.
2.В следующем потоке обработайте его так, чтобы он мог использоваться EgdeTPU.
2-1
python object_detection/export_tflite_ssd_graph.py --pipeline_config_path=object_detection/ssd_mobilenet_v3_large_coco/pipeline.config --trained_checkpoint_prefix=object_detection/test0001/save/model.ckpt-10000 --output_directory=object_detection/test0001/tflite --add_postprocessing_op=true
2-2
tflite_convert --output_file=object_detection/test0001/tflite/test.tflite --graph_def_file=object_detection/test0001/tflite/tflite_graph.pb --inference_type=FLOAT --input_arrays=normalized_input_image_tensor --input_shape=1,320,320,3 --output_arrays=TFLite_Detection_PostProcess,TFLite_Detection_PostProcess:1,TFLite_Detection_PostProcess:2,TFLite_Detection_PostProcess:3 --default_ranges_min=0 --default_ranges_max=255 --mean_values=128 --std_dev_values=127 --inference_type=QUANTIZED_UINT8 --allow_custom_ops
2 -3
edgetpu_compiler test.tflite
log
Edge TPU Compiler version 2.1.302470888
Input: test.tflite
Output: test_edgetpu.tflite
Operator Count Status
LOGISTIC 1 More than one subgraph is not supported
CUSTOM 1 Operation is working on an unsupported data type
ADD 22 More than one subgraph is not supported
MUL 28 More than one subgraph is not supported
CONCATENATION 2 More than one subgraph is not supported
CONV_2D 1 Mapped to Edge TPU
CONV_2D 66 More than one subgraph is not supported
DEPTHWISE_CONV_2D 31 More than one subgraph is not supported
HARD_SWISH 8 Operation not supported
RESHAPE 13 More than one subgraph is not supported
MEAN 8 More than one subgraph is not supported
Это связано с квантованием после тренировки? Я не знаю, какой именно метод квантования c после тренировки.
Секунда
Во время обучения я думал о тренировке с учетом квантования. Я создал следующее для pipe.comfig, но происходит ошибка. 1
python object_detection/model_main.py --pipeline_config_path=”object_detection/ssd_mobilenet_v3_large_coco/pipeline.config" --model_dir="./object_detection/test0001/save" --alsologtostderr
pipe.config
# SSDLite with Mobilenet v3 large feature extractor.
# Trained on COCO14, initialized from scratch.
# 3.22M parameters, 1.02B FLOPs
# TPU-compatible.
# 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 {
inplace_batchnorm_update: true
freeze_batchnorm: false
num_classes: 2
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
use_matmul_gather: true
}
}
similarity_calculator {
iou_similarity {
}
}
encode_background_as_zeros: true
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: 320
width: 320
}
}
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
use_depthwise: true
box_code_size: 4
apply_sigmoid_to_scores: false
class_prediction_bias_init: -4.6
conv_hyperparams {
activation: RELU_6,
regularizer {
l2_regularizer {
weight: 0.00004
}
}
initializer {
random_normal_initializer {
stddev: 0.03
mean: 0.0
}
}
batch_norm {
train: true,
scale: true,
center: true,
decay: 0.97,
epsilon: 0.001,
}
}
}
}
feature_extractor {
type: 'ssd_mobilenet_v3_large'
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: 0.97,
epsilon: 0.001,
}
}
override_base_feature_extractor_hyperparams: true
}
loss {
classification_loss {
weighted_sigmoid_focal {
alpha: 0.75,
gamma: 2.0
}
}
localization_loss {
weighted_smooth_l1 {
delta: 1.0
}
}
classification_weight: 1.0
localization_weight: 1.0
}
normalize_loss_by_num_matches: true
normalize_loc_loss_by_codesize: true
post_processing {
batch_non_max_suppression {
score_threshold: 1e-8
iou_threshold: 0.6
max_detections_per_class: 100
max_total_detections: 100
use_static_shapes: true
}
score_converter: SIGMOID
}
}
}
train_config: {
batch_size: 4
sync_replicas: true
startup_delay_steps: 0
replicas_to_aggregate: 32
num_steps: 10000
data_augmentation_options {
random_horizontal_flip {
}
}
data_augmentation_options {
ssd_random_crop {
}
}
optimizer {
momentum_optimizer: {
learning_rate: {
cosine_decay_learning_rate {
total_steps: 10000
warmup_steps: 10000
}
}
momentum_optimizer_value: 0.9
}
use_moving_average: false
}
fine_tune_checkpoint: "./object_detection/ssd_mobilenet_v3_large_coco/model.ckpt"
from_detection_checkpoint: true
max_number_of_boxes: 100
unpad_groundtruth_tensors: false
}
train_input_reader: {
tf_record_input_reader {
input_path: "./object_detection/test0001/train/test????.tfrecord"
}
label_map_path: "./object_detection/test0001/tf_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: "./object_detection/test0001/val/test????.tfrecord"
}
label_map_path: "./object_detection/test0001/tf_label_map.pbtxt"
shuffle: false
num_readers: 1
}
graph_rewriter {
quantization {
delay: 48000
weight_bits: 8
activation_bits: 8
}
}
log
tensorflow.python.framework.errors_impl.NotFoundError: Restoring from checkpoint failed. This is most likely due to a Variable name or other graph key that is missing from the checkpoint. Please ensure that you have not altered the graph expected based on the checkpoint. Original error:
Key FeatureExtractor/MobilenetV3/Conv/conv_quant/max not found in checkpoint
среда
OS:WINDOWS10
tensorflow:1.15.0rc3