diff --git a/configs/sparse_rcnn/README.md b/configs/sparse_rcnn/README.md
new file mode 100644
index 0000000000000000000000000000000000000000..5443b037f247938fc9a72194fff62c9a27cedc50
--- /dev/null
+++ b/configs/sparse_rcnn/README.md
@@ -0,0 +1,25 @@
+# Sparse R-CNN: End-to-End Object Detection with Learnable Proposals
+
+
+## Introduction
+Sparse RCNN is a purely sparse method for object detection in images.
+
+
+## Model Zoo
+
+| Backbone        | Proposals | lr schedule | Box AP | download   | config |
+| :-------------- | :-----: | :------------: | :-----: | :-----: | :-----: |
+| ResNet50-FPN | 100 | 3x |  43.0  | [download](https://paddledet.bj.bcebos.com/models/sparse_rcnn_r50_fpn_3x_pro100_coco.pdparams) | [config](./sparse_rcnn_r50_fpn_3x_pro100_coco.yml) |
+| ResNet50-FPN | 300 | 3x |  44.6  | [download](https://paddledet.bj.bcebos.com/models/sparse_rcnn_r50_fpn_3x_pro300_coco.pdparams) | [config](./sparse_rcnn_r50_fpn_3x_pro300_coco.yml) |
+
+## Citations
+```
+@misc{sun2021sparse,
+      title={Sparse R-CNN: End-to-End Object Detection with Learnable Proposals},
+      author={Peize Sun and Rufeng Zhang and Yi Jiang and Tao Kong and Chenfeng Xu and Wei Zhan and Masayoshi Tomizuka and Lei Li and Zehuan Yuan and Changhu Wang and Ping Luo},
+      year={2021},
+      eprint={2011.12450},
+      archivePrefix={arXiv},
+      primaryClass={cs.CV}
+}
+```
diff --git a/configs/sparse_rcnn/_base_/optimizer_3x.yml b/configs/sparse_rcnn/_base_/optimizer_3x.yml
new file mode 100644
index 0000000000000000000000000000000000000000..19e1037130158909632a4d6515f6adf53cf5ad3c
--- /dev/null
+++ b/configs/sparse_rcnn/_base_/optimizer_3x.yml
@@ -0,0 +1,17 @@
+epoch: 36
+
+LearningRate:
+  base_lr: 0.000025
+  schedulers:
+  - !PiecewiseDecay
+    gamma: 0.1
+    milestones: [28, 34]
+  - !LinearWarmup
+    start_factor: 0.01
+    steps: 1000
+
+OptimizerBuilder:
+  clip_grad_by_norm: 1.0
+  optimizer:
+    type: AdamW
+    weight_decay: 0.0001
diff --git a/configs/sparse_rcnn/_base_/sparse_rcnn_r50_fpn.yml b/configs/sparse_rcnn/_base_/sparse_rcnn_r50_fpn.yml
new file mode 100644
index 0000000000000000000000000000000000000000..9f7516fcd8652c866ad660f2f0afc9e36f1a6033
--- /dev/null
+++ b/configs/sparse_rcnn/_base_/sparse_rcnn_r50_fpn.yml
@@ -0,0 +1,44 @@
+architecture: SparseRCNN
+pretrain_weights: https://paddledet.bj.bcebos.com/models/pretrained/ResNet50_cos_pretrained.pdparams
+
+SparseRCNN:
+  backbone: ResNet
+  neck: FPN
+  head: SparseRCNNHead
+  postprocess: SparsePostProcess
+
+ResNet:
+  # index 0 stands for res2
+  depth: 50
+  norm_type: bn
+  freeze_at: 0
+  return_idx: [0,1,2,3]
+  num_stages: 4
+
+FPN:
+  out_channel: 256
+
+SparseRCNNHead:
+  head_hidden_dim: 256
+  head_dim_feedforward: 2048
+  nhead: 8
+  head_dropout: 0.0
+  head_cls: 1
+  head_reg: 3
+  head_dim_dynamic: 64
+  head_num_dynamic: 2
+  head_num_heads: 6
+  deep_supervision: true
+  num_proposals: 100
+  loss_func: SparseRCNNLoss
+
+SparseRCNNLoss:
+  losses: ["labels", "boxes"]
+  focal_loss_alpha: 0.25
+  focal_loss_gamma: 2.0
+  class_weight: 2.0
+  l1_weight: 5.0
+  giou_weight: 2.0
+
+SparsePostProcess:
+  num_proposals: 100
diff --git a/configs/sparse_rcnn/_base_/sparse_rcnn_reader.yml b/configs/sparse_rcnn/_base_/sparse_rcnn_reader.yml
new file mode 100644
index 0000000000000000000000000000000000000000..f4226473d78168c407337ee156b25ab7b7462b67
--- /dev/null
+++ b/configs/sparse_rcnn/_base_/sparse_rcnn_reader.yml
@@ -0,0 +1,44 @@
+worker_num: 4
+use_process: true
+
+TrainReader:
+  sample_transforms:
+  - Decode: {}
+  - RandomResize: {target_size: [[480, 1333], [512, 1333], [544, 1333], [576, 1333], [608, 1333], [640, 1333], [672, 1333], [704, 1333], [736, 1333], [768, 1333], [800, 1333]], keep_ratio: true, interp: 1}
+  - RandomFlip: {prob: 0.5}
+  - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
+  - Permute: {}
+  batch_transforms:
+  - PadBatch: {pad_to_stride: 32}
+  - Gt2SparseRCNNTarget: {}
+  batch_size: 4
+  shuffle: true
+  drop_last: true
+  collate_batch: false
+  use_process: true
+
+EvalReader:
+  sample_transforms:
+  - Decode: {}
+  - Resize: {interp: 1, target_size: [800, 1333], keep_ratio: True}
+  - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
+  - Permute: {}
+  batch_transforms:
+  - PadBatch: {pad_to_stride: 32}
+  - Gt2SparseRCNNTarget: {}
+  batch_size: 1
+  shuffle: false
+  drop_last: false
+  use_process: true
+
+TestReader:
+  sample_transforms:
+  - Decode: {}
+  - Resize: {interp: 1, target_size: [800, 1333], keep_ratio: True}
+  - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
+  - Permute: {}
+  batch_transforms:
+  - PadBatch: {pad_to_stride: 32}
+  - Gt2SparseRCNNTarget: {train: false}
+  batch_size: 1
+  shuffle: false
diff --git a/configs/sparse_rcnn/sparse_rcnn_r50_fpn_3x_pro100_coco.yml b/configs/sparse_rcnn/sparse_rcnn_r50_fpn_3x_pro100_coco.yml
new file mode 100644
index 0000000000000000000000000000000000000000..495d6af390c5c71587c76d2a0e44d1d296ee919e
--- /dev/null
+++ b/configs/sparse_rcnn/sparse_rcnn_r50_fpn_3x_pro100_coco.yml
@@ -0,0 +1,10 @@
+_BASE_: [
+  '../datasets/coco_detection.yml',
+  '../runtime.yml',
+  '_base_/sparse_rcnn_r50_fpn.yml',
+  '_base_/optimizer_3x.yml',
+  '_base_/sparse_rcnn_reader.yml',
+]
+
+num_classes: 80
+weights: output/sparse_rcnn_r50b_fpn_pro100/model_final
diff --git a/configs/sparse_rcnn/sparse_rcnn_r50_fpn_3x_pro300_coco.yml b/configs/sparse_rcnn/sparse_rcnn_r50_fpn_3x_pro300_coco.yml
new file mode 100644
index 0000000000000000000000000000000000000000..9268c5ffde14cbbbd5b6336f487e0be3f86bb910
--- /dev/null
+++ b/configs/sparse_rcnn/sparse_rcnn_r50_fpn_3x_pro300_coco.yml
@@ -0,0 +1,19 @@
+_BASE_: [
+  '../datasets/coco_detection.yml',
+  '../runtime.yml',
+  '_base_/sparse_rcnn_r50_fpn.yml',
+  '_base_/optimizer_3x.yml',
+  '_base_/sparse_rcnn_reader.yml',
+]
+
+num_classes: 80
+weights: output/sparse_rcnn_r50b_fpn_pro300/model_final
+
+snapshot_epoch: 1
+
+
+SparseRCNNHead:
+  num_proposals: 300
+
+SparsePostProcess:
+  num_proposals: 300