From b29cd0b545322945a4df3a74596a2bfb504dadd5 Mon Sep 17 00:00:00 2001
From: liuyuan <3463423099@qq.com>
Date: Wed, 15 Mar 2023 17:49:51 +0800
Subject: [PATCH] [Feature]: Add PixMIM config

---
 ...vit-base-p16_8xb512-amp-coslr-300e_in1k.py | 27 +++++++++++++++++++
 mmselfsup/models/algorithms/pixmim.py         |  1 +
 2 files changed, 28 insertions(+)
 create mode 100644 configs/selfsup/pixmim/pixmim_vit-base-p16_8xb512-amp-coslr-300e_in1k.py

diff --git a/configs/selfsup/pixmim/pixmim_vit-base-p16_8xb512-amp-coslr-300e_in1k.py b/configs/selfsup/pixmim/pixmim_vit-base-p16_8xb512-amp-coslr-300e_in1k.py
new file mode 100644
index 00000000..765c779e
--- /dev/null
+++ b/configs/selfsup/pixmim/pixmim_vit-base-p16_8xb512-amp-coslr-300e_in1k.py
@@ -0,0 +1,27 @@
+_base_ = '../mae/mae_vit-base-p16_8xb512-amp-coslr-300e_in1k.py'
+
+train_pipeline = [
+    dict(type='LoadImageFromFile'),
+    dict(type='mmcls.ToPIL', to_rgb=True),
+    dict(type='mmcls.torchvision/Resize', size=224),
+    dict(
+        type='mmcls.torchvision/RandomCrop',
+        size=224,
+        padding=4,
+        padding_mode='reflect'),
+    dict(type='mmcls.torchvision/RandomHorizontalFlip', p=0.5),
+    dict(type='mmcls.ToNumpy', to_rgb=True),
+    dict(type='PackSelfSupInputs', meta_keys=['img_path'])
+]
+
+train_dataloader = dict(dataset=dict(pipeline=train_pipeline))
+
+# model settings
+model = dict(
+    type='PixMIM',
+    target_generator=dict(
+        type='LowFreqTargetGenerator', radius=40, img_size=224),
+)
+
+# randomness
+randomness = dict(seed=2, diff_rank_seed=True)
\ No newline at end of file
diff --git a/mmselfsup/models/algorithms/pixmim.py b/mmselfsup/models/algorithms/pixmim.py
index b134ea3e..6a89fcca 100644
--- a/mmselfsup/models/algorithms/pixmim.py
+++ b/mmselfsup/models/algorithms/pixmim.py
@@ -22,6 +22,7 @@ class PixMIM(MAE):
              data_samples: List[SelfSupDataSample],
              **kwargs) -> Dict[str, torch.Tensor]:
         """The forward function in training.
+
         Args:
             inputs (List[torch.Tensor]): The input images.
             data_samples (List[SelfSupDataSample]): All elements required