| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148 |
- epoch: 12
- use_gpu: true
- use_xpu: false
- use_mlu: false
- use_npu: false
- target_metrics: mask
- log_iter: 20
- save_dir: output
- snapshot_epoch: 1
- print_flops: false
- print_params: false
- # Dataset
- metric: COCO
- num_classes: 80
- worker_num: 8
- TrainDataset:
- name: COCODataSet
- image_dir: train2017
- anno_path: annotations/instances_train2017.json
- dataset_dir: dataset/coco
- data_fields: ['image', 'gt_bbox', 'gt_class', 'gt_poly', 'is_crowd']
- EvalDataset:
- name: COCODataSet
- image_dir: val2017
- anno_path: annotations/instances_val2017.json
- dataset_dir: dataset/coco
- TestDataset:
- name: ImageFolder
- anno_path: annotations/instances_val2017.json # also support txt (like VOC's label_list.txt)
- dataset_dir: dataset/coco # if set, anno_path will be 'dataset_dir/anno_path'
- TrainReader:
- sample_transforms:
- - Decode: {}
- - Poly2Mask: {}
- - Resize: {interp: 1, target_size: [800, 1333], keep_ratio: True}
- - RandomFlip: {}
- - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
- - Permute: {}
- batch_transforms:
- - PadBatch: {pad_to_stride: 32}
- - Gt2Solov2Target: {num_grids: [40, 36, 24, 16, 12],
- scale_ranges: [[1, 96], [48, 192], [96, 384], [192, 768], [384, 2048]],
- coord_sigma: 0.2}
- batch_size: 2
- shuffle: true
- drop_last: true
- EvalReader:
- sample_transforms:
- - Decode: {}
- - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
- - Resize: {interp: 1, target_size: [800, 1333], keep_ratio: True}
- - Permute: {}
- batch_transforms:
- - PadBatch: {pad_to_stride: 32}
- batch_size: 1
- shuffle: false
- drop_last: false
- TestReader:
- sample_transforms:
- - Decode: {}
- - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
- - Resize: {interp: 1, target_size: [800, 1333], keep_ratio: True}
- - Permute: {}
- batch_transforms:
- - PadBatch: {pad_to_stride: 32}
- batch_size: 1
- shuffle: false
- drop_last: false
- LearningRate:
- base_lr: 0.01
- schedulers:
- - !PiecewiseDecay
- gamma: 0.1
- milestones: [8, 11]
- - !LinearWarmup
- start_factor: 0.
- steps: 1000
- OptimizerBuilder:
- optimizer:
- momentum: 0.9
- type: Momentum
- regularizer:
- factor: 0.0001
- type: L2
- # Model
- architecture: SOLOv2
- pretrain_weights: https://paddledet.bj.bcebos.com/models/solov2_r50_fpn_1x_coco.pdparams
- SOLOv2:
- backbone: ResNet
- neck: FPN
- solov2_head: SOLOv2Head
- mask_head: SOLOv2MaskHead
- ResNet:
- depth: 50
- freeze_at: 0
- return_idx: [0,1,2,3]
- num_stages: 4
- FPN:
- out_channel: 256
- SOLOv2Head:
- seg_feat_channels: 512
- stacked_convs: 4
- num_grids: [40, 36, 24, 16, 12]
- kernel_out_channels: 256
- solov2_loss: SOLOv2Loss
- mask_nms: MaskMatrixNMS
- SOLOv2MaskHead:
- mid_channels: 128
- out_channels: 256
- start_level: 0
- end_level: 3
- SOLOv2Loss:
- ins_loss_weight: 3.0
- focal_loss_gamma: 2.0
- focal_loss_alpha: 0.25
- MaskMatrixNMS:
- pre_nms_top_n: 500
- post_nms_top_n: 100
- # Exporting the model
- export:
- post_process: True
- nms: True
- benchmark: False
- fuse_conv_bn: False
|
