add new models for detection (#1850)

* add new models for detection

* add models for detection v2

* remove no-use files

* remove files

README.md

@@ -70,7 +70,7 @@ PaddleX 3.0 覆盖了 16 条产业级模型产线，其中 9 条基础产线可
     <td>通用目标检测</td>
     <td>目标检测</td>
     <td>PicoDet-S<br/>PicoDet-L<details>
-    <summary><b>more</b></summary><br/>PP-YOLOE_plus-S<br/>PP-YOLOE_plus-M<br/>PP-YOLOE_plus-L<br/>PP-YOLOE_plus-X<br/>RT-DETR-L<br/>RT-DETR-H<br/>RT-DETR-X<br/>RT-DETR-R18<br/>RT-DETR-R50</details></td>
+    <summary><b>more</b></summary><br/>PP-YOLOE_plus-S<br/>PP-YOLOE_plus-M<br/>PP-YOLOE_plus-L<br/>PP-YOLOE_plus-X<br/>RT-DETR-L<br/>RT-DETR-H<br/>RT-DETR-X<br/>RT-DETR-R18<br/>RT-DETR-R50<br/>YOLOv3-DarkNet53<br/>YOLOv3-MobileNetV3<br/>YOLOv3-ResNet50_vd_DCN<br/>YOLOX-L<br/>YOLOX-M<br/>YOLOX-N<br/>YOLOX-S<br/>YOLOX-T<br/>YOLOX-X</details></td>
   </tr>
   <tr>
     <td>基础产线</td>

docs/tutorials/models/support_model_list.md

@@ -99,6 +99,21 @@
 | :--- | :---: |
 | PicoDet-S | [PicoDet-S.yaml](../../../paddlex/configs/object_detection/PicoDet-S.yaml)|
 | PicoDet-L | [PicoDet-L.yaml](../../../paddlex/configs/object_detection/PicoDet-L.yaml)|
+### 4. YOLOv3 系列
+| 模型名称 | config |
+| :--- | :---: |
+| YOLOv3-DarkNet53 | [YOLOv3-DarkNet53.yaml](../../../paddlex/configs/object_detection/YOLOv3-DarkNet53.yaml)|
+| YOLOv3-MobileNetV3 | [YOLOv3-MobileNetV3.yaml](../../../paddlex/configs/object_detection/YOLOv3-MobileNetV3.yaml)|
+| YOLOv3-ResNet50_vd_DCN | [YOLOv3-ResNet50_vd_DCN.yaml](../../../paddlex/configs/object_detection/YOLOv3-ResNet50_vd_DCN.yaml)|
+### 5. YOLOX 系列
+| 模型名称 | config |
+| :--- | :---: |
+| YOLOX-L | [YOLOX-L.yaml](../../../paddlex/configs/object_detection/YOLOX-L.yaml)|
+| YOLOX-M | [YOLOX-M.yaml](../../../paddlex/configs/object_detection/YOLOX-M.yaml)|
+| YOLOX-N | [YOLOX-N.yaml](../../../paddlex/configs/object_detection/YOLOX-N.yaml)|
+| YOLOX-S | [YOLOX-S.yaml](../../../paddlex/configs/object_detection/YOLOX-S.yaml)|
+| YOLOX-T | [YOLOX-T.yaml](../../../paddlex/configs/object_detection/YOLOX-T.yaml)|
+| YOLOX-X | [YOLOX-X.yaml](../../../paddlex/configs/object_detection/YOLOX-X.yaml)|
 ## 三、实例分割
 ### 1.Mask-RT-DETR 系列
 | 模型名称 | config |

paddlex/configs/object_detection/YOLOX-L.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOX-L
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.01
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOX-M.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOX-M
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.01
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOX-N.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOX-N
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.01
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOX-S.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOX-S
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.01
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOX-T.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOX-T
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.01
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOX-X.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOX-X
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.01
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOv3-DarkNet53.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOv3-DarkNet53
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 2
+  learning_rate: 0.001
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOv3-MobileNetV3.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOv3-MobileNetV3
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 10
+  batch_size: 8
+  learning_rate: 0.001
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 10
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/configs/object_detection/YOLOv3-ResNet50_vd_DCN.yaml

@@ -0,0 +1,37 @@
+Global:
+  model: YOLOv3-ResNet50_vd_DCN
+  mode: check_dataset # check_dataset/train/evaluate/predict
+  dataset_dir: "/paddle/dataset/paddlex/det/det_coco_examples"
+  device: gpu:0,1,2,3
+  output: "output"
+
+CheckDataset:
+  convert:
+    enable: False
+    src_dataset_type: null
+  split:
+    enable: False
+    train_percent: null
+    val_percent: null
+
+Train:
+  num_classes: 4
+  epochs_iters: 270
+  batch_size: 8
+  learning_rate: 0.001
+  pretrain_weight_path: null
+  warmup_steps: 100
+  resume_path: null
+  log_interval: 20
+  eval_interval: 1
+
+Evaluate:
+  weight_path: "output/best_model.pdparams"
+  log_interval: 10
+
+Predict:
+  model_dir: "output/best_model"
+  input_path: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/general_object_detection_002.png"
+  kernel_option:
+    run_mode: paddle
+    batch_size: 1

paddlex/modules/base/predictor/utils/official_models.py

@@ -146,6 +146,24 @@ CLIP_vit_large_patch14_224_infer.tar",
     "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/RT-DETR-H_infer.tar",
     "RT-DETR-X":
     "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/RT-DETR-X_infer.tar",
+    "YOLOv3-DarkNet53":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOv3-DarkNet53_infer.tar",
+    "YOLOv3-MobileNetV3":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOv3-MobileNetV3_infer.tar",
+    "YOLOv3-ResNet50_vd_DCN":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOv3-ResNet50_vd_DCN_infer.tar",
+    "YOLOX-L":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOX-L_infer.tar",
+    "YOLOX-M":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOX-M_infer.tar",
+    "YOLOX-N":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOX-N_infer.tar",
+    "YOLOX-S":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOX-S_infer.tar",
+    "YOLOX-T":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOX-T_infer.tar",
+    "YOLOX-X":
+    "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/YOLOX-X_infer.tar",
     "RT-DETR-R18":
     "https://paddle-model-ecology.bj.bcebos.com/paddlex/official_inference_model/paddle3.0/RT-DETR-R18_infer.tar",
     "RT-DETR-R50":

paddlex/modules/object_detection/model_list.py

@@ -26,4 +26,13 @@ MODELS = [
     'RT-DETR-R50',
     'RT-DETR-X',
     'PicoDet_layout_1x',
+    "YOLOv3-DarkNet53",
+    "YOLOv3-MobileNetV3",
+    "YOLOv3-ResNet50_vd_DCN",
+    "YOLOX-L",
+    "YOLOX-M",
+    "YOLOX-N",
+    "YOLOX-S",
+    "YOLOX-T",
+    "YOLOX-X",
 ]

paddlex/repo_apis/PaddleDetection_api/configs/YOLOX-L.yaml

@@ -0,0 +1,150 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 100
+save_dir: output
+snapshot_epoch: 10
+print_flops: false
+print_params: false
+
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - Mosaic:
+        prob: 1.0
+        input_dim: [640, 640]
+        degrees: [-10, 10]
+        scale: [0.1, 2.0]
+        shear: [-2, 2]
+        translate: [-0.1, 0.1]
+        enable_mixup: True
+        mixup_prob: 1.0
+        mixup_scale: [0.5, 1.5]
+    - AugmentHSV: {is_bgr: False, hgain: 5, sgain: 30, vgain: 30}
+    - PadResize: {target_size: 640}
+    - RandomFlip: {}
+  batch_transforms:
+    - Permute: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: True
+  collate_batch: False
+  mosaic_epoch: 285
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 4
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 1
+
+
+architecture: YOLOX
+norm_type: sync_bn
+use_ema: True
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+act: silu
+find_unused_parameters: True
+
+depth_mult: 1.0
+width_mult: 1.0
+
+YOLOX:
+  backbone: CSPDarkNet
+  neck: YOLOCSPPAN
+  head: YOLOXHead
+  size_stride: 32
+  size_range: [15, 25] # multi-scale range [480*480 ~ 800*800]
+
+CSPDarkNet:
+  arch: "X"
+  return_idx: [2, 3, 4]
+  depthwise: False
+
+YOLOCSPPAN:
+  depthwise: False
+
+YOLOXHead:
+  l1_epoch: 285
+  depthwise: False
+  loss_weight: {cls: 1.0, obj: 1.0, iou: 5.0, l1: 1.0}
+  assigner:
+    name: SimOTAAssigner
+    candidate_topk: 10
+    use_vfl: False
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 10000
+    keep_top_k: 1000
+    score_threshold: 0.001
+    nms_threshold: 0.65
+    # For speed while keep high mAP, you can modify 'nms_top_k' to 1000 and 'keep_top_k' to 100, the mAP will drop about 0.1%.
+    # For high speed demo, you can modify 'score_threshold' to 0.25 and 'nms_threshold' to 0.45, but the mAP will drop a lot.
+
+
+epoch: 300
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !CosineDecay
+    max_epochs: 300
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 15
+  - !ExpWarmup
+    epochs: 5
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.9
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOX-M.yaml

@@ -0,0 +1,150 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 100
+save_dir: output
+snapshot_epoch: 10
+print_flops: false
+print_params: false
+
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - Mosaic:
+        prob: 1.0
+        input_dim: [640, 640]
+        degrees: [-10, 10]
+        scale: [0.1, 2.0]
+        shear: [-2, 2]
+        translate: [-0.1, 0.1]
+        enable_mixup: True
+        mixup_prob: 1.0
+        mixup_scale: [0.5, 1.5]
+    - AugmentHSV: {is_bgr: False, hgain: 5, sgain: 30, vgain: 30}
+    - PadResize: {target_size: 640}
+    - RandomFlip: {}
+  batch_transforms:
+    - Permute: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: True
+  collate_batch: False
+  mosaic_epoch: 285
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 4
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 1
+
+
+architecture: YOLOX
+norm_type: sync_bn
+use_ema: True
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+act: silu
+find_unused_parameters: True
+
+depth_mult: 0.67
+width_mult: 0.75
+
+YOLOX:
+  backbone: CSPDarkNet
+  neck: YOLOCSPPAN
+  head: YOLOXHead
+  size_stride: 32
+  size_range: [15, 25] # multi-scale range [480*480 ~ 800*800]
+
+CSPDarkNet:
+  arch: "X"
+  return_idx: [2, 3, 4]
+  depthwise: False
+
+YOLOCSPPAN:
+  depthwise: False
+
+YOLOXHead:
+  l1_epoch: 285
+  depthwise: False
+  loss_weight: {cls: 1.0, obj: 1.0, iou: 5.0, l1: 1.0}
+  assigner:
+    name: SimOTAAssigner
+    candidate_topk: 10
+    use_vfl: False
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 10000
+    keep_top_k: 1000
+    score_threshold: 0.001
+    nms_threshold: 0.65
+    # For speed while keep high mAP, you can modify 'nms_top_k' to 1000 and 'keep_top_k' to 100, the mAP will drop about 0.1%.
+    # For high speed demo, you can modify 'score_threshold' to 0.25 and 'nms_threshold' to 0.45, but the mAP will drop a lot.
+
+
+epoch: 300
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !CosineDecay
+    max_epochs: 300
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 15
+  - !ExpWarmup
+    epochs: 5
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.9
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOX-N.yaml

@@ -0,0 +1,145 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 100
+save_dir: output
+snapshot_epoch: 10
+print_flops: false
+print_params: false
+
+depth_mult: 0.33
+width_mult: 0.25
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - Mosaic:
+        prob: 0.5 # 1.0 in YOLOX-tiny/s/m/l/x
+        input_dim: [640, 640]
+        degrees: [-10, 10]
+        scale: [0.5, 1.5] # [0.1, 2.0] in YOLOX-s/m/l/x
+        shear: [-2, 2]
+        translate: [-0.1, 0.1]
+        enable_mixup: False # True in YOLOX-s/m/l/x
+    - AugmentHSV: {is_bgr: False, hgain: 5, sgain: 30, vgain: 30}
+    - PadResize: {target_size: 640}
+    - RandomFlip: {}
+  batch_transforms:
+    - Permute: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: True
+  collate_batch: False
+  mosaic_epoch: 285
+
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [416, 416], keep_ratio: True, interp: 1}
+    - Pad: {size: [416, 416], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 8
+
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 416, 416]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [416, 416], keep_ratio: True, interp: 1}
+    - Pad: {size: [416, 416], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 1
+
+
+architecture: YOLOX
+norm_type: sync_bn
+use_ema: True
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+act: silu
+find_unused_parameters: True
+
+
+YOLOX:
+  backbone: CSPDarkNet
+  neck: YOLOCSPPAN
+  head: YOLOXHead
+  size_stride: 32
+  size_range: [10, 20] # multi-scale range [320*320 ~ 640*640]
+
+CSPDarkNet:
+  arch: "X"
+  return_idx: [2, 3, 4]
+  depthwise: True
+
+YOLOCSPPAN:
+  depthwise: True
+
+YOLOXHead:
+  depthwise: True
+  l1_epoch: 285
+  loss_weight: {cls: 1.0, obj: 1.0, iou: 5.0, l1: 1.0}
+  assigner:
+    name: SimOTAAssigner
+    candidate_topk: 10
+    use_vfl: False
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 10000
+    keep_top_k: 1000
+    score_threshold: 0.001
+    nms_threshold: 0.65
+
+epoch: 300
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !CosineDecay
+    max_epochs: 300
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 15
+  - !ExpWarmup
+    epochs: 5
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.9
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOX-S.yaml

@@ -0,0 +1,150 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 100
+save_dir: output
+snapshot_epoch: 10
+print_flops: false
+print_params: false
+
+depth_mult: 0.33
+width_mult: 0.50
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - Mosaic:
+        prob: 1.0
+        input_dim: [640, 640]
+        degrees: [-10, 10]
+        scale: [0.1, 2.0]
+        shear: [-2, 2]
+        translate: [-0.1, 0.1]
+        enable_mixup: True
+        mixup_prob: 1.0
+        mixup_scale: [0.5, 1.5]
+    - AugmentHSV: {is_bgr: False, hgain: 5, sgain: 30, vgain: 30}
+    - PadResize: {target_size: 640}
+    - RandomFlip: {}
+  batch_transforms:
+    - Permute: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: True
+  collate_batch: False
+  mosaic_epoch: 285
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 4
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 1
+
+
+architecture: YOLOX
+norm_type: sync_bn
+use_ema: True
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+act: silu
+find_unused_parameters: True
+
+
+YOLOX:
+  backbone: CSPDarkNet
+  neck: YOLOCSPPAN
+  head: YOLOXHead
+  size_stride: 32
+  size_range: [15, 25] # multi-scale range [480*480 ~ 800*800]
+
+CSPDarkNet:
+  arch: "X"
+  return_idx: [2, 3, 4]
+  depthwise: False
+
+YOLOCSPPAN:
+  depthwise: False
+
+YOLOXHead:
+  l1_epoch: 285
+  depthwise: False
+  loss_weight: {cls: 1.0, obj: 1.0, iou: 5.0, l1: 1.0}
+  assigner:
+    name: SimOTAAssigner
+    candidate_topk: 10
+    use_vfl: False
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 10000
+    keep_top_k: 1000
+    score_threshold: 0.001
+    nms_threshold: 0.65
+    # For speed while keep high mAP, you can modify 'nms_top_k' to 1000 and 'keep_top_k' to 100, the mAP will drop about 0.1%.
+    # For high speed demo, you can modify 'score_threshold' to 0.25 and 'nms_threshold' to 0.45, but the mAP will drop a lot.
+
+
+epoch: 300
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !CosineDecay
+    max_epochs: 300
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 15
+  - !ExpWarmup
+    epochs: 5
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.9
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOX-T.yaml

@@ -0,0 +1,144 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 100
+save_dir: output
+snapshot_epoch: 10
+print_flops: false
+print_params: false
+
+depth_mult: 0.33
+width_mult: 0.375
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - Mosaic:
+        prob: 1.0
+        input_dim: [640, 640]
+        degrees: [-10, 10]
+        scale: [0.5, 1.5] # [0.1, 2.0] in YOLOX-s/m/l/x
+        shear: [-2, 2]
+        translate: [-0.1, 0.1]
+        enable_mixup: False # True in YOLOX-s/m/l/x
+    - AugmentHSV: {is_bgr: False, hgain: 5, sgain: 30, vgain: 30}
+    - PadResize: {target_size: 640}
+    - RandomFlip: {}
+  batch_transforms:
+    - Permute: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: True
+  collate_batch: False
+  mosaic_epoch: 285
+
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [416, 416], keep_ratio: True, interp: 1}
+    - Pad: {size: [416, 416], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 8
+
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 416, 416]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [416, 416], keep_ratio: True, interp: 1}
+    - Pad: {size: [416, 416], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 1
+
+
+architecture: YOLOX
+norm_type: sync_bn
+use_ema: True
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+act: silu
+find_unused_parameters: True
+
+YOLOX:
+  backbone: CSPDarkNet
+  neck: YOLOCSPPAN
+  head: YOLOXHead
+  size_stride: 32
+  size_range: [10, 20] # multi-scale ragne [320*320 ~ 640*640]
+
+CSPDarkNet:
+  arch: "X"
+  return_idx: [2, 3, 4]
+  depthwise: False
+
+YOLOCSPPAN:
+  depthwise: False
+
+YOLOXHead:
+  l1_epoch: 285
+  depthwise: False
+  loss_weight: {cls: 1.0, obj: 1.0, iou: 5.0, l1: 1.0}
+  assigner:
+    name: SimOTAAssigner
+    candidate_topk: 10
+    use_vfl: False
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 10000
+    keep_top_k: 1000
+    score_threshold: 0.001
+    nms_threshold: 0.65
+
+epoch: 300
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !CosineDecay
+    max_epochs: 300
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 15
+  - !ExpWarmup
+    epochs: 5
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.9
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOX-X.yaml

@@ -0,0 +1,150 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 100
+save_dir: output
+snapshot_epoch: 10
+print_flops: false
+print_params: false
+
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - Mosaic:
+        prob: 1.0
+        input_dim: [640, 640]
+        degrees: [-10, 10]
+        scale: [0.1, 2.0]
+        shear: [-2, 2]
+        translate: [-0.1, 0.1]
+        enable_mixup: True
+        mixup_prob: 1.0
+        mixup_scale: [0.5, 1.5]
+    - AugmentHSV: {is_bgr: False, hgain: 5, sgain: 30, vgain: 30}
+    - PadResize: {target_size: 640}
+    - RandomFlip: {}
+  batch_transforms:
+    - Permute: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: True
+  collate_batch: False
+  mosaic_epoch: 285
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 4
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [640, 640], keep_ratio: True, interp: 1}
+    - Pad: {size: [640, 640], fill_value: [114., 114., 114.]}
+    - Permute: {}
+  batch_size: 1
+
+
+architecture: YOLOX
+norm_type: sync_bn
+use_ema: True
+ema_decay: 0.9999
+ema_decay_type: "exponential"
+act: silu
+find_unused_parameters: True
+
+depth_mult: 1.33
+width_mult: 1.25
+
+YOLOX:
+  backbone: CSPDarkNet
+  neck: YOLOCSPPAN
+  head: YOLOXHead
+  size_stride: 32
+  size_range: [15, 25] # multi-scale range [480*480 ~ 800*800]
+
+CSPDarkNet:
+  arch: "X"
+  return_idx: [2, 3, 4]
+  depthwise: False
+
+YOLOCSPPAN:
+  depthwise: False
+
+YOLOXHead:
+  l1_epoch: 285
+  depthwise: False
+  loss_weight: {cls: 1.0, obj: 1.0, iou: 5.0, l1: 1.0}
+  assigner:
+    name: SimOTAAssigner
+    candidate_topk: 10
+    use_vfl: False
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 10000
+    keep_top_k: 1000
+    score_threshold: 0.001
+    nms_threshold: 0.65
+    # For speed while keep high mAP, you can modify 'nms_top_k' to 1000 and 'keep_top_k' to 100, the mAP will drop about 0.1%.
+    # For high speed demo, you can modify 'score_threshold' to 0.25 and 'nms_threshold' to 0.45, but the mAP will drop a lot.
+
+
+epoch: 300
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !CosineDecay
+    max_epochs: 300
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 15
+  - !ExpWarmup
+    epochs: 5
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.9
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOv3-DarkNet53.yaml

@@ -0,0 +1,152 @@
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 20
+save_dir: output
+snapshot_epoch: 5
+print_flops: false
+print_params: false
+
+
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+
+worker_num: 2
+TrainReader:
+  inputs_def:
+    num_max_boxes: 50
+  sample_transforms:
+    - Decode: {}
+    - RandomDistort: {}
+    - RandomExpand: {fill_value: [123.675, 116.28, 103.53], ratio: 2.0}
+    - RandomCrop: {}
+    - RandomFlip: {}
+  batch_transforms:
+    - BatchRandomResize: {target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608], random_size: True, random_interp: True, keep_ratio: False}
+    - NormalizeBox: {}
+    - PadBox: {num_max_boxes: 50}
+    - BboxXYXY2XYWH: {}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+    - 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], iou_thresh: 0.5}
+  batch_size: 8
+  shuffle: true
+  drop_last: true
+  mixup_epoch: -1
+  use_shared_memory: true
+
+EvalReader:
+  inputs_def:
+    num_max_boxes: 50
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [608, 608], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+  batch_size: 1
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 608, 608]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [608, 608], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+  batch_size: 1
+
+
+
+architecture: YOLOv3
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/DarkNet53_pretrained.pdparams
+norm_type: bn
+
+YOLOv3:
+  backbone: DarkNet
+  neck: YOLOv3FPN
+  yolo_head: YOLOv3Head
+  post_process: BBoxPostProcess
+
+DarkNet:
+  depth: 53
+  return_idx: [2, 3, 4]
+
+# use default config
+# YOLOv3FPN:
+
+YOLOv3Head:
+  anchors: [[10, 13], [16, 30], [33, 23],
+            [30, 61], [62, 45], [59, 119],
+            [116, 90], [156, 198], [373, 326]]
+  anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
+  loss: YOLOv3Loss
+
+YOLOv3Loss:
+  ignore_thresh: 0.5
+  downsample: [32, 16, 8]
+  label_smooth: false
+
+BBoxPostProcess:
+  decode:
+    name: YOLOBox
+    conf_thresh: 0.005
+    downsample_ratio: 32
+    clip_bbox: true
+  nms:
+    name: MultiClassNMS
+    keep_top_k: 100
+    score_threshold: 0.01
+    nms_threshold: 0.45
+    nms_top_k: 1000
+    cpu: True
+
+
+epoch: 270
+
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones:
+    - 216
+    - 243
+  - !LinearWarmup
+    start_factor: 0.
+    steps: 4000
+
+OptimizerBuilder:
+  optimizer:
+    momentum: 0.9
+    type: Momentum
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+# Exporting the model
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOv3-MobileNetV3.yaml

@@ -0,0 +1,153 @@
+# Runtime
+epoch: 270
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 20
+save_dir: output
+snapshot_epoch: 1
+print_flops: false
+print_params: false
+
+# Dataset
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+worker_num: 2
+TrainReader:
+  inputs_def:
+    num_max_boxes: 50
+  sample_transforms:
+    - Decode: {}
+    - Mixup: {alpha: 1.5, beta: 1.5}
+    - RandomDistort: {}
+    - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
+    - RandomCrop: {}
+    - RandomFlip: {}
+  batch_transforms:
+    - BatchRandomResize: {target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608], random_size: True, random_interp: True, keep_ratio: False}
+    - NormalizeBox: {}
+    - PadBox: {num_max_boxes: 50}
+    - BboxXYXY2XYWH: {}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+    - 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
+  drop_last: true
+  mixup_epoch: 250
+  use_shared_memory: true
+
+EvalReader:
+  inputs_def:
+    num_max_boxes: 50
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [608, 608], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+  batch_size: 1
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 608, 608]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [608, 608], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+  batch_size: 1
+
+# Model
+architecture: YOLOv3
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/MobileNetV3_large_x1_0_ssld_pretrained.pdparams
+norm_type: sync_bn
+
+YOLOv3:
+  backbone: MobileNetV3
+  neck: YOLOv3FPN
+  yolo_head: YOLOv3Head
+  post_process: BBoxPostProcess
+
+MobileNetV3:
+  model_name: large
+  scale: 1.
+  with_extra_blocks: false
+  extra_block_filters: []
+  feature_maps: [7, 13, 16]
+
+# use default config
+# YOLOv3FPN:
+
+YOLOv3Head:
+  anchors: [[10, 13], [16, 30], [33, 23],
+            [30, 61], [62, 45], [59, 119],
+            [116, 90], [156, 198], [373, 326]]
+  anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
+  loss: YOLOv3Loss
+
+YOLOv3Loss:
+  ignore_thresh: 0.7
+  downsample: [32, 16, 8]
+  label_smooth: false
+
+BBoxPostProcess:
+  decode:
+    name: YOLOBox
+    conf_thresh: 0.005
+    downsample_ratio: 32
+    clip_bbox: true
+  nms:
+    name: MultiClassNMS
+    keep_top_k: 100
+    score_threshold: 0.01
+    nms_threshold: 0.45
+    nms_top_k: 1000
+
+# Optimizer
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones:
+    - 216
+    - 243
+  - !LinearWarmup
+    start_factor: 0.
+    steps: 4000
+
+OptimizerBuilder:
+  optimizer:
+    momentum: 0.9
+    type: Momentum
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+# Export
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/configs/YOLOv3-ResNet50_vd_DCN.yaml

@@ -0,0 +1,154 @@
+# Runtime
+epoch: 270
+use_gpu: true
+use_xpu: false
+use_mlu: false
+use_npu: false
+log_iter: 20
+save_dir: output
+snapshot_epoch: 1
+print_flops: false
+print_params: false
+
+# Dataset
+metric: COCO
+num_classes: 80
+
+TrainDataset:
+  name: COCODataSet
+  image_dir: train2017
+  anno_path: annotations/instances_train2017.json
+  dataset_dir: dataset/coco
+  data_fields: ['image', 'gt_bbox', 'gt_class', 'is_crowd']
+
+EvalDataset:
+  name: COCODataSet
+  image_dir: val2017
+  anno_path: annotations/instances_val2017.json
+  dataset_dir: dataset/coco
+  allow_empty: true
+
+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'
+
+worker_num: 2
+TrainReader:
+  inputs_def:
+    num_max_boxes: 50
+  sample_transforms:
+    - Decode: {}
+    - Mixup: {alpha: 1.5, beta: 1.5}
+    - RandomDistort: {}
+    - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
+    - RandomCrop: {}
+    - RandomFlip: {}
+  batch_transforms:
+    - BatchRandomResize: {target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608], random_size: True, random_interp: True, keep_ratio: False}
+    - NormalizeBox: {}
+    - PadBox: {num_max_boxes: 50}
+    - BboxXYXY2XYWH: {}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+    - 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
+  drop_last: true
+  mixup_epoch: 250
+  use_shared_memory: true
+
+EvalReader:
+  inputs_def:
+    num_max_boxes: 50
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [608, 608], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+  batch_size: 1
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 608, 608]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: [608, 608], keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
+    - Permute: {}
+  batch_size: 1
+
+# Model
+architecture: YOLOv3
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/ResNet50_vd_ssld_pretrained.pdparams
+norm_type: sync_bn
+
+YOLOv3:
+  backbone: ResNet
+  neck: YOLOv3FPN
+  yolo_head: YOLOv3Head
+  post_process: BBoxPostProcess
+
+ResNet:
+  depth: 50
+  variant: d
+  return_idx: [1, 2, 3]
+  dcn_v2_stages: [3]
+  freeze_at: -1
+  freeze_norm: false
+  norm_decay: 0.
+
+# YOLOv3FPN:
+
+YOLOv3Head:
+  anchors: [[10, 13], [16, 30], [33, 23],
+            [30, 61], [62, 45], [59, 119],
+            [116, 90], [156, 198], [373, 326]]
+  anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
+  loss: YOLOv3Loss
+
+YOLOv3Loss:
+  ignore_thresh: 0.7
+  downsample: [32, 16, 8]
+  label_smooth: false
+
+BBoxPostProcess:
+  decode:
+    name: YOLOBox
+    conf_thresh: 0.005
+    downsample_ratio: 32
+    clip_bbox: true
+  nms:
+    name: MultiClassNMS
+    keep_top_k: 100
+    score_threshold: 0.01
+    nms_threshold: 0.45
+    nms_top_k: 1000
+
+# Optimizer
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones:
+    - 216
+    - 243
+  - !LinearWarmup
+    start_factor: 0.
+    steps: 4000
+
+OptimizerBuilder:
+  optimizer:
+    momentum: 0.9
+    type: Momentum
+  regularizer:
+    factor: 0.0005
+    type: L2
+
+# Export
+export:
+  post_process: True  # Whether post-processing is included in the network when export model.
+  nms: True           # Whether NMS is included in the network when export model.
+  benchmark: False    # It is used to testing model performance, if set `True`, post-process and NMS will not be exported.
+  fuse_conv_bn: False

paddlex/repo_apis/PaddleDetection_api/object_det/register.py

@@ -191,3 +191,120 @@ register_model_info({
         'amp': ['OFF']
     },
 })
+
+register_model_info({
+    'model_name': 'YOLOv3-DarkNet53',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOv3-DarkNet53.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOv3-MobileNetV3',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOv3-MobileNetV3.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOv3-ResNet50_vd_DCN',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOv3-ResNet50_vd_DCN.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOX-L',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOX-L.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOX-M',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOX-M.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOX-N',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOX-N.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOX-S',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOX-S.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOX-T',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOX-T.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})
+
+register_model_info({
+    'model_name': 'YOLOX-X',
+    'suite': 'Det',
+    'config_path': osp.join(PDX_CONFIG_DIR, 'YOLOX-X.yaml'),
+    'supported_apis': ['train', 'evaluate', 'predict', 'export', 'infer'],
+    'supported_dataset_types': ['COCODetDataset'],
+    'supported_train_opts': {
+        'device': ['cpu', 'gpu_nxcx', 'xpu', 'npu', 'mlu'],
+        'dy2st': False,
+        'amp': ['OFF']
+    },
+})