add outstanding pcb config (#1523)

configs/pcb/README.md

 # 印刷电路板（PCB）瑕疵数据集模型
 - 印刷电路板（PCB）瑕疵数据集：[数据下载链接](http://robotics.pkusz.edu.cn/resources/dataset/)，是一个公共的合成PCB数据集，由北京大学发布，其中包含1386张图像以及6种缺陷（缺失孔，鼠标咬伤，开路，短路，杂散，伪铜），用于检测，分类和配准任务。我们选取了其中适用与检测任务的693张图像，随机选择593张图像作为训练集，100张图像作为验证集。AIStudio数据集链接：[印刷电路板（PCB）瑕疵数据集](https://aistudio.baidu.com/aistudio/datasetdetail/52914)
 
-## 已提供的基线模型
-- YOLOv3-Darknet
-- Faster-RCNN-R34
-- FCOS-R50-FPN
-
-## 优化后高mAP模型
-将在9月29日发布,敬请期待。
+## 模型
+| 模型  | mAP(Iou=0.50:0.95)           | mAP(Iou=0.50)          | 配置文件 |
+| :-------: | :-------: | :----: | :-----: |
+| Faster-RCNN-R50_vd_FPN_3x | 52.7  | 97.7  | faster_rcnn_r50_vd_fpn_3x.yml |
+| YOLOv3_darknet | 44.8  | 94.6 | yolov3_darknet.yml |
+| FCOS_R50_FPN_multiscale_3x | 54.9 | 98.5 | fcos_r50_fpn_multiscale_3x.yml |

configs/pcb/faster_rcnn_r34_fpn_1x.yml

-architecture: FasterRCNN
-max_iters: 3558
-use_gpu: true
-snapshot_iter: 1000
-log_smooth_window: 20
-save_dir: output
-pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet34_pretrained.tar
-metric: COCO
-weights: output/faster_rcnn_r34_fpn_1x/model_final
-num_classes: 7
-
-FasterRCNN:
-  backbone: ResNet
-  fpn: FPN
-  rpn_head: FPNRPNHead
-  roi_extractor: FPNRoIAlign
-  bbox_head: BBoxHead
-  bbox_assigner: BBoxAssigner
-
-ResNet:
-  norm_type: bn
-  norm_decay: 0.
-  depth: 34
-  feature_maps: [2, 3, 4, 5]
-  freeze_at: 2
-
-FPN:
-  min_level: 2
-  max_level: 6
-  num_chan: 256
-  spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
-
-FPNRPNHead:
-  anchor_generator:
-    anchor_sizes: [32, 64, 128, 256, 512]
-    aspect_ratios: [0.5, 1.0, 2.0]
-    stride: [16.0, 16.0]
-    variance: [1.0, 1.0, 1.0, 1.0]
-  anchor_start_size: 32
-  min_level: 2
-  max_level: 6
-  num_chan: 256
-  rpn_target_assign:
-    rpn_batch_size_per_im: 256
-    rpn_fg_fraction: 0.5
-    rpn_positive_overlap: 0.7
-    rpn_negative_overlap: 0.3
-    rpn_straddle_thresh: 0.0
-  train_proposal:
-    min_size: 0.0
-    nms_thresh: 0.7
-    pre_nms_top_n: 2000
-    post_nms_top_n: 2000
-  test_proposal:
-    min_size: 0.0
-    nms_thresh: 0.7
-    pre_nms_top_n: 1000
-    post_nms_top_n: 1000
-
-FPNRoIAlign:
-  canconical_level: 4
-  canonical_size: 224
-  min_level: 2
-  max_level: 5
-  box_resolution: 7
-  sampling_ratio: 2
-
-BBoxAssigner:
-  batch_size_per_im: 512
-  bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
-  bg_thresh_lo: 0.0
-  bg_thresh_hi: 0.5
-  fg_fraction: 0.25
-  fg_thresh: 0.5
-
-BBoxHead:
-  head: TwoFCHead
-  nms:
-    keep_top_k: 100
-    nms_threshold: 0.5
-    score_threshold: 0.05
-
-TwoFCHead:
-  mlp_dim: 1024
-
-LearningRate:
-  base_lr: 0.0025
-  schedulers:
-  - !PiecewiseDecay
-    gamma: 0.1
-    milestones: [2372, 3261]
-  - !LinearWarmup
-    start_factor: 0.1
-    steps: 500
-
-OptimizerBuilder:
-  optimizer:
-    momentum: 0.9
-    type: Momentum
-  regularizer:
-    factor: 0.0001
-    type: L2
-
-_READER_: 'faster_fpn_reader.yml'
-TrainReader:
-  batch_size: 2

configs/pcb/faster_fpn_reader.yml → configs/pcb/faster_rcnn_r50_vd_fpn_3x.yml

+architecture: FasterRCNN
+max_iters: 11000
+snapshot_iter: 1000
+use_gpu: true
+log_smooth_window: 20
+save_dir: output
+pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar
+weights: output/faster_rcnn_r50_vd_fpn_2x/model_final
+metric: COCO
+num_classes: 7
+
+FasterRCNN:
+  backbone: ResNet
+  fpn: FPN
+  rpn_head: FPNRPNHead
+  roi_extractor: FPNRoIAlign
+  bbox_head: BBoxHead
+  bbox_assigner: BBoxAssigner
+
+ResNet:
+  depth: 50
+  feature_maps: [2, 3, 4, 5]
+  freeze_at: 2
+  norm_type: bn
+  variant: d
+
+FPN:
+  max_level: 6
+  min_level: 2
+  num_chan: 256
+  spatial_scale: [0.03125, 0.0625, 0.125, 0.25]
+
+FPNRPNHead:
+  anchor_generator:
+    anchor_sizes: [8, 16, 32, 64, 128]
+    aspect_ratios: [0.5, 1.0, 2.0]
+    stride: [16.0, 16.0]
+    variance: [1.0, 1.0, 1.0, 1.0]
+  anchor_start_size: 8
+  max_level: 6
+  min_level: 2
+  num_chan: 256
+  rpn_target_assign:
+    rpn_batch_size_per_im: 256
+    rpn_fg_fraction: 0.5
+    rpn_negative_overlap: 0.3
+    rpn_positive_overlap: 0.7
+    rpn_straddle_thresh: 0.0
+  train_proposal:
+    min_size: 0.0
+    nms_thresh: 0.7
+    post_nms_top_n: 2000
+    pre_nms_top_n: 2000
+  test_proposal:
+    min_size: 0.0
+    nms_thresh: 0.7
+    post_nms_top_n: 1000
+    pre_nms_top_n: 1000
+
+FPNRoIAlign:
+  canconical_level: 4
+  canonical_size: 224
+  max_level: 5
+  min_level: 2
+  box_resolution: 7
+  sampling_ratio: 2
+
+BBoxAssigner:
+  batch_size_per_im: 512
+  bbox_reg_weights: [0.1, 0.1, 0.2, 0.2]
+  bg_thresh_hi: 0.5
+  bg_thresh_lo: 0.0
+  fg_fraction: 0.25
+  fg_thresh: 0.5
+
+BBoxHead:
+  head: TwoFCHead
+  nms:
+    keep_top_k: 100
+    nms_threshold: 0.5
+    score_threshold: 0.05
+
+TwoFCHead:
+  mlp_dim: 1024
+
+LearningRate:
+  base_lr: 0.0025
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones: [7122, 9800]
+  - !LinearWarmup
+    start_factor: 0.333
+    steps: 500
+
+OptimizerBuilder:
+  optimizer:
+    momentum: 0.9
+    type: Momentum
+  regularizer:
+    factor: 0.0001
+    type: L2
+
 TrainReader:
   inputs_def:
     fields: ['image', 'im_info', 'im_id', 'gt_bbox', 'gt_class', 'is_crowd']
@@ -28,7 +131,7 @@ TrainReader:
   - !PadBatch
     pad_to_stride: 32
     use_padded_im_info: false
-  batch_size: 1
+  batch_size: 2
   shuffle: true
   worker_num: 2
   use_process: false

configs/pcb/fcos_r50_fpn_1x.yml → configs/pcb/fcos_r50_fpn_multiscale_3x.yml

 architecture: FCOS
-max_iters: 3558
+max_iters: 11000
 use_gpu: true
 snapshot_iter: 1000
 log_smooth_window: 20
 log_iter: 20
 save_dir: output
-pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_cos_pretrained.tar
+pretrain_weights: https://paddle-imagenet-models-name.bj.bcebos.com/ResNet50_vd_ssld_pretrained.tar
 metric: COCO
-weights: output/fcos_r50_fpn_1x/model_final
+weights: output/fcos_r50_fpn_multiscale_2x/model_final
 num_classes: 6
 
 FCOS:
@@ -16,11 +16,13 @@ FCOS:
   fcos_head: FCOSHead
 
 ResNet:
-  norm_type: affine_channel
+  norm_type: bn
   norm_decay: 0.
   depth: 50
   feature_maps: [3, 4, 5]
   freeze_at: 2
+  dcn_v2_stages: [3, 4, 5]
+  variant: d
 
 FPN:
   min_level: 3
@@ -38,7 +40,7 @@ FCOSHead:
   fcos_loss: FCOSLoss
   norm_reg_targets: True
   centerness_on_reg: True
-  use_dcn_in_tower: False
+  use_dcn_in_tower: True
   nms: MultiClassNMS
 
 MultiClassNMS:
@@ -59,7 +61,7 @@ LearningRate:
   schedulers:
   - !PiecewiseDecay
     gamma: 0.1
-    milestones: [2372, 3261]
+    milestones: [7122, 9800]
   - !LinearWarmup
     start_factor: 0.3333333333333333
     steps: 500
@@ -92,7 +94,7 @@ TrainReader:
     mean: [0.485,0.456,0.406]
     std: [0.229, 0.224,0.225]
   - !ResizeImage
-    target_size: 800
+    target_size: [640, 672, 704, 736, 768, 800]
     max_size: 1333
     interp: 1
     use_cv2: true

configs/pcb/yolov3_darknet.yml

 architecture: YOLOv3
 use_gpu: true
-max_iters: 1780
+max_iters: 11000
 log_smooth_window: 20
 save_dir: output
 snapshot_iter: 1000
@@ -21,9 +21,9 @@ DarkNet:
 
 YOLOv3Head:
   anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
-  anchors: [[10, 13], [16, 30], [33, 23],
-            [30, 61], [62, 45], [59, 119],
-            [116, 90], [156, 198], [373, 326]]
+  anchors: [[12, 22], [19, 20], [30, 20],
+            [22, 28], [16, 41], [45, 22],
+            [26, 43], [36, 34], [53, 53]]
   norm_decay: 0.
   yolo_loss: YOLOv3Loss
   nms:
@@ -35,11 +35,6 @@ YOLOv3Head:
     score_threshold: 0.01
 
 YOLOv3Loss:
-  # batch_size here is only used for fine grained loss, not used
-  # for training batch_size setting, training batch_size setting
-  # is in configs/yolov3_reader.yml TrainReader.batch_size, batch
-  # size here should be set as same value as TrainReader.batch_size
-  batch_size: 8
   ignore_thresh: 0.7
   label_smooth: true
 
@@ -49,8 +44,8 @@ LearningRate:
   - !PiecewiseDecay
     gamma: 0.1
     milestones:
-    - 1180
-    - 1630
+    - 7122
+    - 9800
   - !LinearWarmup
     start_factor: 0.
     steps: 500
@@ -63,4 +58,102 @@ OptimizerBuilder:
     factor: 0.0005
     type: L2
 
-_READER_: 'yolov3_reader.yml'
+TrainReader:
+  inputs_def:
+    fields: ['image', 'gt_bbox', 'gt_class', 'gt_score']
+    num_max_boxes: 50
+  dataset:
+    !COCODataSet
+      image_dir: images
+      anno_path: Annotations/train.json
+      dataset_dir: /home/aistudio/work/PCB_DATASET/
+      with_background: false
+  sample_transforms:
+    - !DecodeImage
+      to_rgb: True
+    - !NormalizeBox {}
+    - !PadBox
+      num_max_boxes: 50
+    - !BboxXYXY2XYWH {}
+  batch_transforms:
+  - !RandomShape
+    sizes: [640, 704, 768, 832, 896, 960]
+    random_inter: True
+  - !NormalizeImage
+    mean: [0.485, 0.456, 0.406]
+    std: [0.229, 0.224, 0.225]
+    is_scale: True
+    is_channel_first: false
+  - !Permute
+    to_bgr: false
+    channel_first: True
+  # Gt2YoloTarget is only used when use_fine_grained_loss set as true,
+  # this operator will be deleted automatically if use_fine_grained_loss
+  # is set as false
+  - !Gt2YoloTarget
+    anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
+    anchors: [[12, 22], [19, 20], [30, 20],
+            [22, 28], [16, 41], [45, 22],
+            [26, 43], [36, 34], [53, 53]]
+    downsample_ratios: [32, 16, 8]
+  batch_size: 4
+  shuffle: true
+  drop_last: true
+  worker_num: 8
+  bufsize: 4
+  use_process: true
+
+EvalReader:
+  inputs_def:
+    fields: ['image', 'im_size', 'im_id']
+    num_max_boxes: 50
+  dataset:
+    !COCODataSet
+      image_dir: images
+      anno_path: Annotations/val.json
+      dataset_dir: /home/aistudio/work/PCB_DATASET/
+      with_background: false
+  sample_transforms:
+    - !DecodeImage
+      to_rgb: True
+    - !ResizeImage
+      target_size: 896
+      interp: 2
+    - !NormalizeImage
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+      is_scale: True
+      is_channel_first: false
+    - !PadBox
+      num_max_boxes: 50
+    - !Permute
+      to_bgr: false
+      channel_first: True
+  batch_size: 2
+  drop_empty: false
+  worker_num: 8
+  bufsize: 4
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 896, 896]
+    fields: ['image', 'im_size', 'im_id']
+  dataset:
+    !ImageFolder
+      anno_path: /home/aistudio/work/PCB_DATASET/Annotations/val.json
+      with_background: false
+  sample_transforms:
+    - !DecodeImage
+      to_rgb: True
+    - !ResizeImage
+      target_size: 896
+      interp: 2
+    - !NormalizeImage
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+      is_scale: True
+      is_channel_first: false
+    - !Permute
+      to_bgr: false
+      channel_first: True
+  batch_size: 1

configs/pcb/yolov3_reader.yml

-TrainReader:
-  inputs_def:
-    fields: ['image', 'gt_bbox', 'gt_class', 'gt_score']
-    num_max_boxes: 50
-  dataset:
-    !COCODataSet
-      image_dir: images
-      anno_path: Annotations/train.json
-      dataset_dir: /home/aistudio/work/PCB_DATASET
-      with_background: false
-  sample_transforms:
-    - !DecodeImage
-      to_rgb: True
-      with_mixup: True
-    - !MixupImage
-      alpha: 1.5
-      beta: 1.5
-    - !ColorDistort {}
-    - !RandomExpand
-      fill_value: [123.675, 116.28, 103.53]
-    - !RandomCrop {}
-    - !RandomFlipImage
-      is_normalized: false
-    - !NormalizeBox {}
-    - !PadBox
-      num_max_boxes: 50
-    - !BboxXYXY2XYWH {}
-  batch_transforms:
-  - !RandomShape
-    sizes: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
-    random_inter: True
-  - !NormalizeImage
-    mean: [0.485, 0.456, 0.406]
-    std: [0.229, 0.224, 0.225]
-    is_scale: True
-    is_channel_first: false
-  - !Permute
-    to_bgr: false
-    channel_first: True
-  # Gt2YoloTarget is only used when use_fine_grained_loss set as true,
-  # this operator will be deleted automatically if use_fine_grained_loss
-  # is set as false
-  - !Gt2YoloTarget
-    anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
-    anchors: [[10, 13], [16, 30], [33, 23],
-              [30, 61], [62, 45], [59, 119],
-              [116, 90], [156, 198], [373, 326]]
-    downsample_ratios: [32, 16, 8]
-  batch_size: 8
-  shuffle: true
-  mixup_epoch: 250
-  drop_last: true
-  worker_num: 8
-  bufsize: 16
-  use_process: true
-
-
-EvalReader:
-  inputs_def:
-    fields: ['image', 'im_size', 'im_id']
-    num_max_boxes: 50
-  dataset:
-    !COCODataSet
-      image_dir: images
-      anno_path: Annotations/val.json
-      dataset_dir: /home/aistudio/work/PCB_DATASET
-      with_background: false
-  sample_transforms:
-    - !DecodeImage
-      to_rgb: True
-    - !ResizeImage
-      target_size: 608
-      interp: 2
-    - !NormalizeImage
-      mean: [0.485, 0.456, 0.406]
-      std: [0.229, 0.224, 0.225]
-      is_scale: True
-      is_channel_first: false
-    - !PadBox
-      num_max_boxes: 50
-    - !Permute
-      to_bgr: false
-      channel_first: True
-  batch_size: 8
-  drop_empty: false
-  worker_num: 8
-  bufsize: 16
-
-TestReader:
-  inputs_def:
-    image_shape: [3, 608, 608]
-    fields: ['image', 'im_size', 'im_id']
-  dataset:
-    !ImageFolder
-      anno_path: /home/aistudio/work/PCB_DATASET/Annotations/val.json
-      with_background: false
-  sample_transforms:
-    - !DecodeImage
-      to_rgb: True
-    - !ResizeImage
-      target_size: 608
-      interp: 2
-    - !NormalizeImage
-      mean: [0.485, 0.456, 0.406]
-      std: [0.229, 0.224, 0.225]
-      is_scale: True
-      is_channel_first: false
-    - !Permute
-      to_bgr: false
-      channel_first: True
-  batch_size: 1