first init

configs/runtime.yml

+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
+
+# 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

configs/vitdet/README.md

+# Vision Transformer Detection
+
+## Introduction
+
+- [Context Autoencoder for Self-Supervised Representation Learning](https://arxiv.org/abs/2202.03026)  
+- [Benchmarking Detection Transfer Learning with Vision Transformers](https://arxiv.org/pdf/2111.11429.pdf)  
+
+Object detection is a central downstream task used to
+test if pre-trained network parameters confer benefits, such
+as improved accuracy or training speed. The complexity
+of object detection methods can make this benchmarking
+non-trivial when new architectures, such as Vision Transformer (ViT) models, arrive.
+
+## Model Zoo
+
+| Model | Backbone | Pretrained | Scheduler | Images/GPU  | Box AP | Mask AP | Config | Download |
+|:------:|:--------:|:--------------:|:--------------:|:--------------:|:--------------:|:------:|:------:|:--------:|
+| PP-YOLOE | ViT-base | CAE | 36e | 2 | 52.2 | - | [config](./ppyoloe_vit_base_csppan_cae_36e_coco.yml) | [model](https://bj.bcebos.com/v1/paddledet/models/ppyoloe_vit_base_csppan_cae_36e_coco.pdparams) |
+
+
+**Notes:**
+- Model is trained on COCO train2017 dataset and evaluated on val2017 results of `mAP(IoU=0.5:0.95)
+- Base model is trained on 8x32G V100 GPU, large model on 8x80G A100
+
+## Citations
+```
+@article{chen2022context,
+  title={Context autoencoder for self-supervised representation learning},
+  author={Chen, Xiaokang and Ding, Mingyu and Wang, Xiaodi and Xin, Ying and Mo, Shentong and Wang, Yunhao and Han, Shumin and Luo, Ping and Zeng, Gang and Wang, Jingdong},
+  journal={arXiv preprint arXiv:2202.03026},
+  year={2022}
+}
+
+@article{DBLP:journals/corr/abs-2111-11429,
+  author    = {Yanghao Li and
+               Saining Xie and
+               Xinlei Chen and
+               Piotr Doll{\'{a}}r and
+               Kaiming He and
+               Ross B. Girshick},
+  title     = {Benchmarking Detection Transfer Learning with Vision Transformers},
+  journal   = {CoRR},
+  volume    = {abs/2111.11429},
+  year      = {2021},
+  url       = {https://arxiv.org/abs/2111.11429},
+  eprinttype = {arXiv},
+  eprint    = {2111.11429},
+  timestamp = {Fri, 26 Nov 2021 13:48:43 +0100},
+  biburl    = {https://dblp.org/rec/journals/corr/abs-2111-11429.bib},
+  bibsource = {dblp computer science bibliography, https://dblp.org}
+}
+
+@article{Cai_2019,
+   title={Cascade R-CNN: High Quality Object Detection and Instance Segmentation},
+   ISSN={1939-3539},
+   url={http://dx.doi.org/10.1109/tpami.2019.2956516},
+   DOI={10.1109/tpami.2019.2956516},
+   journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
+   publisher={Institute of Electrical and Electronics Engineers (IEEE)},
+   author={Cai, Zhaowei and Vasconcelos, Nuno},
+   year={2019},
+   pages={1–1}
+}
+```

configs/vitdet/_base_/optimizer_base_36e.yml

+
+epoch: 36
+
+LearningRate:
+  base_lr: 0.0001
+  schedulers:
+    - !CosineDecay
+      max_epochs: 36
+      min_lr_ratio: 0.1
+    - !LinearWarmup
+      start_factor: 0.001
+      epochs: 1
+
+
+OptimizerBuilder:
+  clip_grad_by_norm: 0.1
+  regularizer: false
+  optimizer:
+    type: AdamW
+    weight_decay: 0.0001

configs/vitdet/_base_/ppyoloe_reader.yml

+worker_num: 4
+eval_height: &eval_height 640
+eval_width: &eval_width 640
+eval_size: &eval_size [*eval_height, *eval_width]
+
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - 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, 640, 672, 704, 736, 768], random_size: True, random_interp: True, keep_ratio: False}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+    - PadGT: {}
+  batch_size: 2
+  shuffle: true
+  drop_last: true
+  use_shared_memory: true
+  collate_batch: true
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *eval_size, keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 2
+
+TestReader:
+  inputs_def:
+    image_shape: [3, *eval_height, *eval_width]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *eval_size, keep_ratio: False, interp: 2}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1

configs/vitdet/ppyoloe_vit_base_csppan_cae_36e_coco.yml

+
+_BASE_: [
+  '../datasets/coco_detection.yml',
+  '../runtime.yml',
+  './_base_/ppyoloe_reader.yml',
+  './_base_/optimizer_base_36e.yml'
+]
+
+weights: output/ppyoloe_vit_base_csppan_cae_36e_coco/model_final
+
+
+snapshot_epoch: 2
+log_iter: 100
+
+
+use_ema: true
+ema_decay: 0.9999
+ema_skip_names: ['yolo_head.proj_conv.weight', 'backbone.pos_embed']
+custom_black_list: ['reduce_mean']
+use_fused_allreduce_gradients: &use_checkpoint False
+
+
+architecture: YOLOv3
+norm_type: sync_bn
+
+YOLOv3:
+  backbone: VisionTransformer
+  neck: YOLOCSPPAN
+  yolo_head: PPYOLOEHead
+  post_process: ~
+
+VisionTransformer:
+  patch_size: 16
+  embed_dim: 768
+  depth: 12
+  num_heads: 12
+  mlp_ratio: 4
+  qkv_bias: True
+  drop_rate: 0.0
+  drop_path_rate: 0.2
+  init_values: 0.1
+  final_norm: False
+  use_rel_pos_bias: False
+  use_sincos_pos_emb: True
+  epsilon: 0.000001 # 1e-6
+  out_indices: [11, ]
+  with_fpn: True
+  num_fpn_levels: 3
+  out_with_norm: False
+  use_checkpoint: *use_checkpoint
+  pretrained: https://bj.bcebos.com/v1/paddledet/models/pretrained/vit_base_cae_pretrained.pdparams
+
+YOLOCSPPAN:
+  in_channels: [768, 768, 768]
+  act: 'silu'
+
+PPYOLOEHead:
+  fpn_strides: [8, 16, 32]
+  in_channels: [768, 768, 768]
+  static_assigner_epoch: -1
+  grid_cell_scale: 5.0
+  grid_cell_offset: 0.5
+  use_varifocal_loss: True
+  loss_weight: {class: 1.0, iou: 2.5, dfl: 0.5}
+  static_assigner:
+    name: ATSSAssigner
+    topk: 9
+  assigner:
+    name: TaskAlignedAssigner
+    topk: 13
+    alpha: 1.0
+    beta: 6.0
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 1000
+    keep_top_k: 300
+    score_threshold: 0.01
+    nms_threshold: 0.7

configs/voc/README.md

+# YOLO on VOC
+
+## 模型库
+
+| 网络模型        | 输入尺寸   | 图片数/GPU | 学习率策略 | TRT-FP16-Latency(ms) | mAP(0.50,11point) | Params(M) | FLOPs(G) |    下载链接       | 配置文件 |
+| :-----------: | :-------: | :-------: | :------: | :------------: | :---------------: | :------------------: |:-----------------: | :------: | :------: |
+| YOLOv5-s        |  640     |    16     |   60e    |     3.2   |  80.3 |  7.24  | 16.54 | [下载链接](https://paddledet.bj.bcebos.com/models/yolov5_s_60e_voc.pdparams) | [配置文件](./yolov5_s_60e_voc.yml) |
+| YOLOv7-tiny     |  640     |    32     |   60e    |     2.6   |  80.2 |  6.23  | 6.90 | [下载链接](https://paddledet.bj.bcebos.com/models/yolov7_tiny_60e_voc.pdparams) | [配置文件](./yolov7_tiny_60e_voc.yml) |
+| YOLOX-s         |  640     |    8      |   40e    |     3.0   |  82.9 |  9.0   |  26.8 | [下载链接](https://paddledet.bj.bcebos.com/models/yolox_s_40e_voc.pdparams) | [配置文件](./yolox_s_40e_voc.yml) |
+| PP-YOLOE+_s     |  640     |    8      |   30e    |     2.9   |  86.7 |  7.93  |  17.36 | [下载链接](https://paddledet.bj.bcebos.com/models/ppyoloe_plus_crn_s_30e_voc.pdparams) | [配置文件](./ppyoloe_plus_crn_s_30e_voc.yml) |
+
+
+**注意:**
+  - 所有YOLO模型均使用VOC数据集训练，mAP为`mAP(IoU=0.5)`的结果，且评估未使用`multi_label`等trick；
+  - 所有YOLO模型均加载各自模型的COCO权重作为预训练，各个配置文件的配置均为默认使用8卡GPU，可作为自定义数据集设置参考，具体精度会因数据集而异；
+  - YOLO检测模型建议**总`batch_size`至少大于`64`**去训练，如果资源不够请**换小模型**或**减小模型的输入尺度**，为了保障较高检测精度，**尽量不要尝试单卡训和总`batch_size`小于`64`训**；
+  - Params(M)和FLOPs(G)均为训练时所测，YOLOv7没有s模型，故选用tiny模型；
+  - TRT-FP16-Latency(ms)测速相关请查看各YOLO模型的config的主页；
+
+
+## 使用教程
+
+### 下载数据集：
+
+下载PaddleDetection团队整理的VOC数据，并放置于`PaddleDetection/dataset/voc`
+```
+wget https://bj.bcebos.com/v1/paddledet/data/voc.zip
+```
+
+### 训练评估预测：
+
+```
+model_name=voc
+job_name=ppyoloe_plus_crn_s_30e_voc # 可修改，如 yolov7_tiny_60e_voc
+
+config=configs/${model_name}/${job_name}.yml
+log_dir=log_dir/${job_name}
+# weights=https://bj.bcebos.com/v1/paddledet/models/${job_name}.pdparams
+weights=output/${job_name}/model_final.pdparams
+
+# 1.训练（单卡/多卡）
+# CUDA_VISIBLE_DEVICES=0 python tools/train.py -c ${config} --eval --amp
+python -m paddle.distributed.launch --log_dir=${log_dir} --gpus 0,1,2,3,4,5,6,7 tools/train.py -c ${config} --eval --amp
+
+# 2.评估
+CUDA_VISIBLE_DEVICES=0 python tools/eval.py -c ${config} -o weights=${weights} --classwise
+
+# 3.预测
+CUDA_VISIBLE_DEVICES=0 python tools/infer.py -c ${config} -o weights=${weights} --infer_img=demo/000000014439_640x640.jpg --draw_threshold=0.5
+```

configs/voc/ppyoloe_plus_crn_s_30e_voc.yml

+_BASE_: [
+  '../ppyoloe/ppyoloe_plus_crn_s_80e_coco.yml',
+  '../datasets/voc.yml',
+]
+log_iter: 50
+snapshot_epoch: 5
+weights: output/ppyoloe_plus_crn_s_30e_voc/model_final
+
+pretrain_weights: https://bj.bcebos.com/v1/paddledet/models/ppyoloe_plus_crn_s_80e_coco.pdparams
+depth_mult: 0.33
+width_mult: 0.50
+
+
+TrainReader:
+  batch_size: 8 # default 8 gpus, total bs = 64
+
+EvalReader:
+  batch_size: 4
+
+
+epoch: 30
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+    - !CosineDecay
+      max_epochs: 36
+    - !LinearWarmup
+      start_factor: 0.
+      epochs: 1
+
+
+PPYOLOEHead:
+  static_assigner_epoch: -1
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 1000
+    keep_top_k: 300
+    score_threshold: 0.01
+    nms_threshold: 0.7

configs/voc/yolov5_s_60e_voc.yml

+_BASE_: [
+  '../yolov5/yolov5_s_300e_coco.yml',
+  '../datasets/voc.yml',
+]
+log_iter: 50
+snapshot_epoch: 5
+weights: output/yolov5_s_60e_voc/model_final
+
+pretrain_weights: https://paddledet.bj.bcebos.com/models/yolov5_s_300e_coco.pdparams
+depth_mult: 0.33
+width_mult: 0.50
+
+
+TrainReader:
+  batch_size: 16 # default 8 gpus, total bs = 128
+
+EvalReader:
+  batch_size: 4
+
+
+epoch: 60
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !YOLOv5LRDecay
+    max_epochs: 60
+    min_lr_ratio: 0.01
+  - !ExpWarmup
+    epochs: 1

configs/voc/yolov7_tiny_60e_voc.yml

+_BASE_: [
+  '../yolov7/yolov7_tiny_300e_coco.yml',
+  '../datasets/voc.yml',
+]
+log_iter: 50
+snapshot_epoch: 5
+weights: output/yolov7_tiny_60e_voc/model_final
+
+pretrain_weights: https://paddledet.bj.bcebos.com/models/yolov7_tiny_300e_coco.pdparams
+arch: tiny
+act: LeakyReLU
+
+
+TrainReader:
+  batch_size: 32 # default 8 gpus, total bs = 256
+
+EvalReader:
+  batch_size: 4
+
+
+epoch: 60
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !YOLOv5LRDecay
+    max_epochs: 60
+    min_lr_ratio: 0.1
+  - !ExpWarmup
+    epochs: 1

configs/voc/yolox_s_40e_voc.yml

+_BASE_: [
+  '../yolox/yolox_s_300e_coco.yml',
+  '../datasets/voc.yml',
+]
+log_iter: 50
+snapshot_epoch: 5
+weights: output/yolox_s_40e_voc/model_final
+
+pretrain_weights: https://paddledet.bj.bcebos.com/models/yolox_s_300e_coco.pdparams
+depth_mult: 0.33
+width_mult: 0.50
+
+TrainReader:
+  batch_size: 8 # default 8 gpus, total bs = 64
+
+EvalReader:
+  batch_size: 4
+
+
+epoch: 40
+LearningRate:
+  base_lr: 0.001
+  schedulers:
+  - !CosineDecay
+    max_epochs: 40
+    min_lr_ratio: 0.05
+    last_plateau_epochs: 4
+  - !ExpWarmup
+    epochs: 1

configs/yolov8/_base_/optimizer_500e.yml

+epoch: 500
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !YOLOv5LRDecay
+    max_epochs: 500
+    min_lr_ratio: 0.01
+  - !ExpWarmup
+    epochs: 5 #3
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.937
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+  clip_grad_by_value: 10.

configs/yolov8/_base_/optimizer_500e_high.yml

+epoch: 500
+
+LearningRate:
+  base_lr: 0.01
+  schedulers:
+  - !YOLOv5LRDecay
+    max_epochs: 500
+    min_lr_ratio: 0.1 #
+  - !ExpWarmup
+    epochs: 5 #3
+
+OptimizerBuilder:
+  optimizer:
+    type: Momentum
+    momentum: 0.937
+    use_nesterov: True
+  regularizer:
+    factor: 0.0005
+    type: L2
+  clip_grad_by_value: 10.

configs/yolov8/_base_/yolov8_cspdarknet.yml

+architecture: YOLOv8
+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 # default: L version
+width_mult: 1.0
+
+YOLOv8:
+  backbone: YOLOv8CSPDarkNet
+  neck: YOLOv8CSPPAN
+  yolo_head: YOLOv8Head
+  post_process: ~
+
+YOLOv8CSPDarkNet:
+  arch: 'P5'
+  return_idx: [2, 3, 4]
+  last_stage_ch: 1024
+  last2_stage_ch: 512
+
+# use default config
+# YOLOv8CSPPAN:
+
+YOLOv8Head:
+  fpn_strides: [8, 16, 32]
+  loss_weight: {class: 0.5, iou: 7.5, dfl: 1.5}
+  assigner:
+    name: TaskAlignedAssigner
+    topk: 10
+    alpha: 0.5
+    beta: 6.0
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 3000
+    keep_top_k: 300
+    score_threshold: 0.001
+    nms_threshold: 0.7

configs/yolov8/_base_/yolov8_reader.yml

+input_height: &input_height 640
+input_width: &input_width 640
+input_size: &input_size [*input_height, *input_width]
+mosaic_epoch: &mosaic_epoch 490 # last 10 epochs close mosaic, totally 500 epochs as default
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - MosaicPerspective: {mosaic_prob: 1.0, boxes_normed: False, target_size: *input_size}
+    - RandomHSV: {hgain: 0.015, sgain: 0.7, vgain: 0.4}
+    - RandomFlip: {}
+  batch_transforms:
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+    - PadGT: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: False
+  use_shared_memory: True
+  collate_batch: True
+  mosaic_epoch: *mosaic_epoch
+
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *input_size, keep_ratio: True, interp: 1}
+    - Pad: {size: *input_size, fill_value: [114., 114., 114.]}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 8
+
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *input_size, keep_ratio: True, interp: 1}
+    - Pad: {size: *input_size, fill_value: [114., 114., 114.]}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+  fuse_normalize: False

configs/yolov8/_base_/yolov8_reader_high_aug.yml

+input_height: &input_height 640
+input_width: &input_width 640
+input_size: &input_size [*input_height, *input_width]
+mosaic_epoch: &mosaic_epoch 490 # last 10 epochs close mosaic, totally 500 epochs as default
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - MosaicPerspective: {mosaic_prob: 1.0, boxes_normed: False, target_size: *input_size, scale: 0.9, mixup_prob: 0.1, copy_paste_prob: 0.1}
+    - RandomHSV: {hgain: 0.015, sgain: 0.7, vgain: 0.4}
+    - RandomFlip: {}
+  batch_transforms:
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+    - PadGT: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: False
+  use_shared_memory: True
+  collate_batch: True
+  mosaic_epoch: *mosaic_epoch
+
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *input_size, keep_ratio: True, interp: 1}
+    - Pad: {size: *input_size, fill_value: [114., 114., 114.]}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 640, 640]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *input_size, keep_ratio: True, interp: 1}
+    - Pad: {size: *input_size, fill_value: [114., 114., 114.]}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+  fuse_normalize: False

configs/yolov8/_base_/yolov8p6_cspdarknet.yml

+architecture: YOLOv8
+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 # default: L version
+width_mult: 1.0
+
+YOLOv8:
+  backbone: YOLOv8CSPDarkNet
+  neck: YOLOv8CSPPANP6
+  yolo_head: YOLOv8Head
+  post_process: ~
+
+YOLOv8CSPDarkNet:
+  arch: 'P6'
+  return_idx: [2, 3, 4, 5]
+  last_stage_ch: 1024
+  last2_stage_ch: 768
+
+# use default config
+# YOLOv8CSPPANP6:
+
+YOLOv8Head:
+  fpn_strides: [8, 16, 32, 64]
+  loss_weight: {class: 0.5, iou: 7.5, dfl: 1.5}
+  assigner:
+    name: TaskAlignedAssigner
+    topk: 10
+    alpha: 0.5
+    beta: 6.0
+  nms:
+    name: MultiClassNMS
+    nms_top_k: 1000
+    keep_top_k: 300
+    score_threshold: 0.001
+    nms_threshold: 0.7

configs/yolov8/_base_/yolov8p6_reader.yml

+input_height: &input_height 1280
+input_width: &input_width 1280
+input_size: &input_size [*input_height, *input_width]
+mosaic_epoch: &mosaic_epoch 490 # last 10 epochs close mosaic, totally 500 epochs as default
+
+worker_num: 4
+TrainReader:
+  sample_transforms:
+    - Decode: {}
+    - MosaicPerspective: {mosaic_prob: 1.0, boxes_normed: False, target_size: *input_size, scale: 0.9, mixup_prob: 0.1, copy_paste_prob: 0.1}
+    - RandomHSV: {hgain: 0.015, sgain: 0.7, vgain: 0.4}
+    - RandomFlip: {}
+  batch_transforms:
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+    - PadGT: {}
+  batch_size: 8
+  shuffle: True
+  drop_last: False
+  use_shared_memory: True
+  collate_batch: True
+  mosaic_epoch: *mosaic_epoch
+
+
+EvalReader:
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *input_size, keep_ratio: True, interp: 1}
+    - Pad: {size: *input_size, fill_value: [114., 114., 114.]}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+
+
+TestReader:
+  inputs_def:
+    image_shape: [3, 1280, 1280]
+  sample_transforms:
+    - Decode: {}
+    - Resize: {target_size: *input_size, keep_ratio: True, interp: 1}
+    - Pad: {size: *input_size, fill_value: [114., 114., 114.]}
+    - NormalizeImage: {mean: [0., 0., 0.], std: [1., 1., 1.], norm_type: none}
+    - Permute: {}
+  batch_size: 1
+  fuse_normalize: False

configs/yolov8/openimagev7/yolov8_l_100e_oiv7.yml

+_BASE_: [
+  '../../datasets/openimagesv7_detection.yml',
+  '../../runtime.yml',
+  '../_base_/optimizer_100e.yml',
+  '../_base_/yolov8_cspdarknet.yml',
+  '../_base_/yolov8_reader_high_aug.yml',
+]
+depth_mult: 1.0
+width_mult: 1.0
+
+log_iter: 50
+snapshot_epoch: 10
+weights: output/yolov8_l_100e_oiv7/model_final
+
+
+YOLOv8CSPDarkNet:
+  last_stage_ch: 512 # The actual channel is int(512 * width_mult), not int(1024 * width_mult) as in YOLOv5
+
+YOLOv8Head:
+  customized_c3: 256
+
+
+TrainReader:
+  batch_size: 16 # default 8 gpus, total bs = 128

configs/yolov8/openimagev7/yolov8_m_100e_oiv7.yml

+_BASE_: [
+  '../../datasets/openimagesv7_detection.yml',
+  '../../runtime.yml',
+  '../_base_/optimizer_100e.yml',
+  '../_base_/yolov8_cspdarknet.yml',
+  '../_base_/yolov8_reader_high_aug.yml',
+]
+depth_mult: 0.67
+width_mult: 0.75
+
+log_iter: 50
+snapshot_epoch: 10
+weights: output/yolov8_m_100e_oiv7/model_final
+
+
+YOLOv8CSPDarkNet:
+  last_stage_ch: 768 # The actual channel is int(768 * width_mult), not int(1024 * width_mult) as in YOLOv5
+
+YOLOv8Head:
+  customized_c3: 192
+
+
+TrainReader:
+  batch_size: 16 # default 8 gpus, total bs = 128