#7 Added configs for tuning on custom dataset

README.md

@@ -397,7 +397,12 @@ CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --master_port=7777 --nproc_per_node=4 trai
 CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --master_port=7777 --nproc_per_node=4 train.py -c configs/dfine/custom/dfine_hgnetv2_${model}_custom.yml --test-only -r model.pth
 ```
 
-4. [Optional] Modify Class Mappings:
+4. Tuning on Custom Dataset
+```shell
+CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --master_port=7777 --nproc_per_node=4 train.py -c configs/dfine/custom/dfine_hgnetv2_${model}_obj2custom.yml --use-amp --seed=0 -t model.pth
+```
+
+5. [Optional] Modify Class Mappings:
 
 When using the Objects365 pre-trained weights to train on your custom dataset, the example assumes that your dataset only contains the classes `'Person'` and `'Car'`. For faster convergence, you can modify `self.obj365_ids` in `src/solver/_solver.py` as follows:
 

README_cn.md

@@ -392,7 +392,12 @@ CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --master_port=7777 --nproc_per_node=4 trai
 CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --master_port=7777 --nproc_per_node=4 train.py -c configs/dfine/custom/dfine_hgnetv2_${model}_custom.yml --test-only -r model.pth
 ```
 
-4. [可选] 修改类映射:
+4. 在自定义数据集上微调
+```shell
+CUDA_VISIBLE_DEVICES=0,1,2,3 torchrun --master_port=7777 --nproc_per_node=4 train.py -c configs/dfine/custom/dfine_hgnetv2_${model}_obj2custom.yml --use-amp --seed=0 -t model.pth
+```
+
+5. [可选] 修改类映射:
 
 在使用 Objects365 预训练权重训练自定义数据集时，示例中假设自定义数据集仅有 `'Person'` 和 `'Car'` 类，您可以将其替换为数据集中对应的任何类别。为了加快收敛，可以在 `src/solver/_solver.py` 中修改 `self.obj365_ids`，如下所示：
 

configs/dfine/custom/objects365/dfine_hgnetv2_l_obj2custom.yml

@@ -0,0 +1,52 @@
+__include__: [
+  '../../dataset/custom_detection.yml',
+  '../../runtime.yml',
+  '../include/dataloader.yml',
+  '../include/optimizer.yml',
+  '../include/dfine_hgnetv2.yml',
+]
+
+output_dir: ./output/dfine_hgnetv2_l_obj2custom
+
+
+DFINE:
+  backbone: HGNetv2
+
+HGNetv2:
+  name: 'B4'
+  return_idx: [1, 2, 3]
+  freeze_stem_only: True
+  freeze_at: 0
+  freeze_norm: True
+
+optimizer:
+  type: AdamW
+  params: 
+    - 
+      params: '^(?=.*backbone)(?!.*norm|bn).*$'
+      lr: 0.0000125
+    - 
+      params: '^(?=.*(?:encoder|decoder))(?=.*(?:norm|bn)).*$'
+      weight_decay: 0.
+
+  lr: 0.00025
+  betas: [0.9, 0.999]
+  weight_decay: 0.000125
+
+
+epoches: 36 # Early stop
+train_dataloader: 
+  dataset: 
+    transforms:
+      policy:
+        epoch: 30
+  collate_fn:
+    stop_epoch: 30
+    ema_restart_decay: 0.9999
+    scale_ori_repeat: 4
+
+ema:
+  warmups: 0
+
+lr_warmup_scheduler:
+  warmup_duration: 0

configs/dfine/custom/objects365/dfine_hgnetv2_m_obj2custom.yml

@@ -0,0 +1,65 @@
+__include__: [
+  '../../dataset/custom_detection.yml',
+  '../../runtime.yml',
+  '../include/dataloader.yml',
+  '../include/optimizer.yml',
+  '../include/dfine_hgnetv2.yml',
+]
+
+output_dir: ./output/dfine_hgnetv2_m_obj2custom
+
+
+DFINE:
+  backbone: HGNetv2
+
+HGNetv2:
+  name: 'B2'
+  return_idx: [1, 2, 3]
+  freeze_at: -1
+  freeze_norm: False
+  use_lab: True
+
+DFINETransformer:
+  num_layers: 4  # 5 6
+  eval_idx: -1  # -2 -3
+
+HybridEncoder:
+  in_channels: [384, 768, 1536]
+  hidden_dim: 256
+  depth_mult: 0.67
+
+optimizer:
+  type: AdamW
+  params: 
+    - 
+      params: '^(?=.*backbone)(?!.*norm|bn).*$'
+      lr: 0.000025
+    - 
+      params: '^(?=.*backbone)(?=.*norm|bn).*$'
+      lr: 0.000025
+      weight_decay: 0.
+    - 
+      params: '^(?=.*(?:encoder|decoder))(?=.*(?:norm|bn|bias)).*$'
+      weight_decay: 0.
+
+  lr: 0.00025
+  betas: [0.9, 0.999]
+  weight_decay: 0.000125
+
+
+epoches: 56 # Early stop
+train_dataloader: 
+  dataset: 
+    transforms:
+      policy:
+        epoch: 48
+  collate_fn:
+    stop_epoch: 48
+    ema_restart_decay: 0.9999
+    scale_ori_repeat: 6
+
+ema:
+  warmups: 0
+
+lr_warmup_scheduler:
+  warmup_duration: 0

configs/dfine/custom/objects365/dfine_hgnetv2_s_obj2custom.yml

@@ -0,0 +1,66 @@
+__include__: [
+  '../../dataset/custom_detection.yml',
+  '../../runtime.yml',
+  '../include/dataloader.yml',
+  '../include/optimizer.yml',
+  '../include/dfine_hgnetv2.yml',
+]
+
+output_dir: ./output/dfine_hgnetv2_s_obj2custom
+
+
+DFINE:
+  backbone: HGNetv2
+
+HGNetv2:
+  name: 'B0'
+  return_idx: [1, 2, 3]
+  freeze_at: -1
+  freeze_norm: False
+  use_lab: True
+
+DFINETransformer:
+  num_layers: 3  # 4 5 6
+  eval_idx: -1  # -2 -3 -4
+
+HybridEncoder:
+  in_channels: [256, 512, 1024]
+  hidden_dim: 256
+  depth_mult: 0.34
+  expansion: 0.5
+
+optimizer:
+  type: AdamW
+  params: 
+    - 
+      params: '^(?=.*backbone)(?!.*norm|bn).*$'
+      lr: 0.000125
+    - 
+      params: '^(?=.*backbone)(?=.*norm|bn).*$'
+      lr: 0.000125
+      weight_decay: 0.
+    - 
+      params: '^(?=.*(?:encoder|decoder))(?=.*(?:norm|bn|bias)).*$'
+      weight_decay: 0.
+
+  lr: 0.00025
+  betas: [0.9, 0.999]
+  weight_decay: 0.000125
+
+
+epoches: 64 # Early stop
+train_dataloader: 
+  dataset: 
+    transforms:
+      policy:
+        epoch: 56
+  collate_fn:
+    stop_epoch: 56
+    ema_restart_decay: 0.9999
+    scale_ori_repeat: 10
+
+ema:
+  warmups: 0
+
+lr_warmup_scheduler:
+  warmup_duration: 0

configs/dfine/custom/objects365/dfine_hgnetv2_x_obj2custom.yml

@@ -0,0 +1,61 @@
+__include__: [
+  '../../dataset/custom_detection.yml',
+  '../../runtime.yml',
+  '../include/dataloader.yml',
+  '../include/optimizer.yml',
+  '../include/dfine_hgnetv2.yml',
+]
+
+output_dir: ./output/dfine_hgnetv2_x_obj2custom
+
+
+DFINE:
+  backbone: HGNetv2
+
+HGNetv2:
+  name: 'B5'
+  return_idx: [1, 2, 3]
+  freeze_stem_only: True
+  freeze_at: 0
+  freeze_norm: True
+
+HybridEncoder:
+  # intra
+  hidden_dim: 384
+  dim_feedforward: 2048
+
+DFINETransformer:
+  feat_channels: [384, 384, 384]
+  reg_scale: 8
+
+optimizer:
+  type: AdamW
+  params: 
+    - 
+      params: '^(?=.*backbone)(?!.*norm|bn).*$'
+      lr: 0.0000025
+    - 
+      params: '^(?=.*(?:encoder|decoder))(?=.*(?:norm|bn)).*$'
+      weight_decay: 0.
+
+  lr: 0.00025
+  betas: [0.9, 0.999]
+  weight_decay: 0.000125
+
+
+epoches: 36 # Early stop
+train_dataloader: 
+  dataset: 
+    transforms:
+      policy:
+        epoch: 30
+  collate_fn:
+    stop_epoch: 30
+    ema_restart_decay: 0.9999
+    scale_ori_repeat: 3
+
+ema:
+  warmups: 0
+
+lr_warmup_scheduler:
+  warmup_duration: 0