PulseFocusPlatform/static/configs/efficientdet_d0.yml

158 lines
3.0 KiB
YAML

architecture: EfficientDet
max_iters: 281250
use_gpu: true
pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/EfficientNetB0_pretrained.tar
weights: output/efficientdet_d0/model_final
log_iter: 20
snapshot_iter: 10000
metric: COCO
save_dir: output
num_classes: 81
use_ema: true
ema_decay: 0.9998
EfficientDet:
backbone: EfficientNet
fpn: BiFPN
efficient_head: EfficientHead
anchor_grid: AnchorGrid
box_loss_weight: 50.
EfficientNet:
# norm_type: sync_bn
# TODO
norm_type: bn
scale: b0
use_se: true
BiFPN:
num_chan: 64
repeat: 3
levels: 5
EfficientHead:
repeat: 3
num_chan: 64
prior_prob: 0.01
num_anchors: 9
gamma: 1.5
alpha: 0.25
delta: 0.1
output_decoder:
score_thresh: 0.05 # originally 0.
nms_thresh: 0.5
pre_nms_top_n: 1000 # originally 5000
detections_per_im: 100
nms_eta: 1.0
AnchorGrid:
anchor_base_scale: 4
num_scales: 3
aspect_ratios: [[1, 1], [1.4, 0.7], [0.7, 1.4]]
LearningRate:
base_lr: 0.16
schedulers:
- !CosineDecayWithSkip
total_steps: 281250
skip_steps: 938
- !LinearWarmup
start_factor: 0.05
steps: 938
OptimizerBuilder:
clip_grad_by_norm: 10.
optimizer:
momentum: 0.9
type: Momentum
regularizer:
factor: 0.00004
type: L2
TrainReader:
inputs_def:
fields: ['image', 'im_id', 'fg_num', 'gt_label', 'gt_target']
dataset:
!COCODataSet
image_dir: train2017
anno_path: annotations/instances_train2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: true
- !RandomFlipImage
prob: 0.5
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !RandomScaledCrop
target_dim: 512
scale_range: [.1, 2.]
interp: 1
- !Permute
to_bgr: false
channel_first: true
- !TargetAssign
image_size: 512
batch_size: 16
shuffle: true
worker_num: 32
bufsize: 16
use_process: true
drop_empty: false
EvalReader:
inputs_def:
fields: ['image', 'im_info', 'im_id']
dataset:
!COCODataSet
image_dir: val2017
anno_path: annotations/instances_val2017.json
dataset_dir: dataset/coco
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeAndPad
target_dim: 512
interp: 1
- !Permute
channel_first: true
to_bgr: false
drop_empty: false
batch_size: 16
shuffle: false
worker_num: 2
TestReader:
inputs_def:
fields: ['image', 'im_info', 'im_id']
image_shape: [3, 512, 512]
dataset:
!ImageFolder
anno_path: annotations/instances_val2017.json
sample_transforms:
- !DecodeImage
to_rgb: true
with_mixup: false
- !NormalizeImage
is_channel_first: false
is_scale: true
mean: [0.485,0.456,0.406]
std: [0.229, 0.224,0.225]
- !ResizeAndPad
target_dim: 512
interp: 1
- !Permute
channel_first: true
to_bgr: false
batch_size: 16
shuffle: false