mindspore/model_zoo/official/cv/maskrcnn_mobilenetv1/default_config.yaml

210 lines
5.9 KiB
YAML

# Builtin Configurations(DO NOT CHANGE THESE CONFIGURATIONS unless you know exactly what you are doing)
enable_modelarts: False
data_url: ""
train_url: ""
checkpoint_url: ""
data_path: "/cache/data"
output_path: "/cache/train"
load_path: "/cache/checkpoint_path"
checkpoint_file: './checkpoint/mask_rcnn-12_7393.ckpt'
device_target: Ascend
enable_profiling: False
# ==============================================================================
modelarts_dataset_unzip_name: 'cocodataset'
need_modelarts_dataset_unzip: True
# MaskRcnn training
only_create_dataset: False
run_distribute: False
do_train: True
do_eval: False
dataset: "coco"
pre_trained: './maskrcnn_mobilenetv1/maskrcnn_mobilenetv1_12_7393.ckpt'
device_id: 0
device_num: 1
rank_id: 0
# MaskRcnn evaluation
ann_file: "./annotations/instances_val2017.json"
checkpoint_path: './maskrcnn_mobilenetv1/maskrcnn_mobilenetv1_12_7393.ckpt'
# maskrcnn mobilnetv1 export"
batch_size_export: 1
ckpt_file: '/cache/data/cocodataset/checkpoint/mask_rcnn-12_7393.ckpt' #############
file_name: "maskrcnn_mobilenetv1"
file_format: "MINDIR"
# maskrcnn-mobilenetv1 inference
img_path: '' # "image file path."
result_path: '' # "result file path."
# ==============================================================================
# config
img_width: 1280
img_height: 768
keep_ratio: True
flip_ratio: 0.5
expand_ratio: 1.0
max_instance_count: 128
mask_shape: (28, 28)
# anchor
feature_shapes: [(192, 320), (96, 160), (48, 80), (24, 40), (12, 20)]
anchor_scales: [8]
anchor_ratios: [0.5, 1.0, 2.0]
anchor_strides: [4, 8, 16, 32, 64]
num_anchors: 3
# fpn
fpn_in_channels: [128, 256, 512, 1024]
fpn_out_channels: 256
fpn_num_outs: 5
# rpn
rpn_in_channels: 256
rpn_feat_channels: 256
rpn_loss_cls_weight: 1.0
rpn_loss_reg_weight: 1.0
rpn_cls_out_channels: 1
rpn_target_means: [0., 0., 0., 0.]
rpn_target_stds: [1.0, 1.0, 1.0, 1.0]
# bbox_assign_sampler
neg_iou_thr: 0.3
pos_iou_thr: 0.7
min_pos_iou: 0.3
num_bboxes: 245520
num_gts: 128
num_expected_neg: 256
num_expected_pos: 128
# proposal
activate_num_classes: 2
use_sigmoid_cls: True
# roi_align
roi_layer: dict(type='RoIAlign', out_size=7, mask_out_size=14, sample_num=2)
roi_align_out_channels: 256
roi_align_featmap_strides: [4, 8, 16, 32]
roi_align_finest_scale: 56
roi_sample_num: 640
# bbox_assign_sampler_stage2
neg_iou_thr_stage2: 0.5
pos_iou_thr_stage2: 0.5
min_pos_iou_stage2: 0.5
num_bboxes_stage2: 2000
num_expected_pos_stage2: 128
num_expected_neg_stage2: 512
num_expected_total_stage2: 512
# rcnn
rcnn_num_layers: 2
rcnn_in_channels: 256
rcnn_fc_out_channels: 1024
rcnn_mask_out_channels: 256
rcnn_loss_cls_weight: 1
rcnn_loss_reg_weight: 1
rcnn_loss_mask_fb_weight: 1
rcnn_target_means: [0., 0., 0., 0.]
rcnn_target_stds: [0.1, 0.1, 0.2, 0.2]
# train proposal
rpn_proposal_nms_across_levels: False
rpn_proposal_nms_pre: 2000
rpn_proposal_nms_post: 2000
rpn_proposal_max_num: 2000
rpn_proposal_nms_thr: 0.7
rpn_proposal_min_bbox_size: 0
# test proposal
rpn_nms_across_levels: False
rpn_nms_pre: 1000
rpn_nms_post: 1000
rpn_max_num: 1000
rpn_nms_thr: 0.7
rpn_min_bbox_min_size: 0
test_score_thr: 0.05
test_iou_thr: 0.5
test_max_per_img: 100
test_batch_size: 2
rpn_head_use_sigmoid: True
rpn_head_weight: 1.0
mask_thr_binary: 0.5
# LR
base_lr: 0.04
base_step: 58633
total_epoch: 13
warmup_step: 500
warmup_ratio: 1/3.0
sgd_momentum: 0.9
# train
batch_size: 2
loss_scale: 1
momentum: 0.91
weight_decay: 0.0001 # 1e-4
pretrain_epoch_size: 0
epoch_size: 12
save_checkpoint: True
save_checkpoint_epochs: 12
keep_checkpoint_max: 12
save_checkpoint_path: "./"
mindrecord_dir: "./MindRecord_COCO"
coco_root: "/cache/data"
train_data_type: "train2017"
val_data_type: "val2017"
instance_set: "annotations/instances_{}.json"
coco_classes: ('background', 'person', 'bicycle', 'car', 'motorcycle', 'airplane', 'bus',
'train', 'truck', 'boat', 'traffic light', 'fire hydrant',
'stop sign', 'parking meter', 'bench', 'bird', 'cat', 'dog',
'horse', 'sheep', 'cow', 'elephant', 'bear', 'zebra',
'giraffe', 'backpack', 'umbrella', 'handbag', 'tie',
'suitcase', 'frisbee', 'skis', 'snowboard', 'sports ball',
'kite', 'baseball bat', 'baseball glove', 'skateboard',
'surfboard', 'tennis racket', 'bottle', 'wine glass', 'cup',
'fork', 'knife', 'spoon', 'bowl', 'banana', 'apple',
'sandwich', 'orange', 'broccoli', 'carrot', 'hot dog', 'pizza',
'donut', 'cake', 'chair', 'couch', 'potted plant', 'bed',
'dining table', 'toilet', 'tv', 'laptop', 'mouse', 'remote',
'keyboard', 'cell phone', 'microwave', 'oven', 'toaster', 'sink',
'refrigerator', 'book', 'clock', 'vase', 'scissors',
'teddy bear', 'hair drier', 'toothbrush')
num_classes: 81
---
# Config description for each option
enable_modelarts: 'Whether training on modelarts, default: False'
data_url: 'Dataset url for obs'
train_url: 'Training output url for obs'
data_path: 'Dataset path for local'
output_path: 'Training output path for local'
ann_file: 'Ann file, default is val.json.'
dataset: "Dataset, default is coco."
checkpoint_path: "Checkpoint file path."
file_name: "output file name."
device_target: 'Target device type'
enable_profiling: 'Whether enable profiling while training, default: False'
only_create_dataset: 'If set it true, only create Mindrecord, default is false.'
run_distribute: 'Run distribute, default is false.'
do_train: 'Do train or not, default is true.'
do_eval: 'Do eval or not, default is false.'
pre_trained: 'Pretrain file path.'
device_id: 'Device id, default is 0.'
device_num: 'Use device nums, default is 1.'
rank_id: 'Rank id, default is 0.'
file_format: 'file format'
img_path: "image file path."
result_path: "result file path."
---
device_target: ['Ascend', 'GPU', 'CPU']
file_format: ["AIR", "ONNX", "MINDIR"]