diff --git a/configs/keypoint/README.md b/configs/keypoint/README.md index a63ac27c2b70e78ac95ec50e35097c8d7aa34551..128a6f0bec26275b3c60e739dc9172408db3c65a 100644 --- a/configs/keypoint/README.md +++ b/configs/keypoint/README.md @@ -23,6 +23,7 @@ COCO数据集 | HRNet-w32 | 384x288 | 77.8 | [hrnet_w32_384x288.pdparams](https://paddledet.bj.bcebos.com/models/keypoint/hrnet_w32_384x288.pdparams) | [config](./hrnet/hrnet_w32_384x288.yml) | | HRNet-w32+DarkPose | 256x192 | 78.0 | [dark_hrnet_w32_256x192.pdparams](https://paddledet.bj.bcebos.com/models/keypoint/dark_hrnet_w32_256x192.pdparams) | [config](./hrnet/dark_hrnet_w32_256x192.yml) | | HRNet-w32+DarkPose | 384x288 | 78.3 | [dark_hrnet_w32_384x288.pdparams](https://paddledet.bj.bcebos.com/models/keypoint/dark_hrnet_w32_384x288.pdparams) | [config](./hrnet/dark_hrnet_w32_384x288.yml) | +| WiderNaiveHRNet-18 | 256x192 | 67.6(+DARK 68.4) | [wider_naive_hrnet_18_256x192_coco.pdparams](https://bj.bcebos.com/v1/paddledet/models/keypoint/wider_naive_hrnet_18_256x192_coco.pdparams) | [config](./lite_hrnet/wider_naive_hrnet_18_256x192_coco.yml) | | LiteHRNet-18 | 256x192 | 66.5 | [lite_hrnet_18_256x192_coco.pdparams](https://bj.bcebos.com/v1/paddledet/models/keypoint/lite_hrnet_18_256x192_coco.pdparams) | [config](./lite_hrnet/lite_hrnet_18_256x192_coco.yml) | | LiteHRNet-18 | 384x288 | 69.7 | [lite_hrnet_18_384x288_coco.pdparams](https://bj.bcebos.com/v1/paddledet/models/keypoint/lite_hrnet_18_384x288_coco.pdparams) | [config](./lite_hrnet/lite_hrnet_18_384x288_coco.yml) | | LiteHRNet-30 | 256x192 | 69.4 | [lite_hrnet_30_256x192_coco.pdparams](https://bj.bcebos.com/v1/paddledet/models/keypoint/lite_hrnet_30_256x192_coco.pdparams) | [config](./lite_hrnet/lite_hrnet_30_256x192_coco.yml) | diff --git a/configs/keypoint/lite_hrnet/wider_naive_hrnet_18_256x192_coco.yml b/configs/keypoint/lite_hrnet/wider_naive_hrnet_18_256x192_coco.yml new file mode 100644 index 0000000000000000000000000000000000000000..a80d08c1fc98e49b8197c1fea1144cdf1efe34ac --- /dev/null +++ b/configs/keypoint/lite_hrnet/wider_naive_hrnet_18_256x192_coco.yml @@ -0,0 +1,140 @@ +use_gpu: true +log_iter: 5 +save_dir: output +snapshot_epoch: 10 +weights: output/wider_naive_hrnet_18_256x192_coco/model_final +epoch: 210 +num_joints: &num_joints 17 +pixel_std: &pixel_std 200 +metric: KeyPointTopDownCOCOEval +num_classes: 1 +train_height: &train_height 256 +train_width: &train_width 192 +trainsize: &trainsize [*train_width, *train_height] +hmsize: &hmsize [48, 64] +flip_perm: &flip_perm [[1, 2], [3, 4], [5, 6], [7, 8], [9, 10], [11, 12], [13, 14], [15, 16]] + + +#####model +architecture: TopDownHRNet + +TopDownHRNet: + backbone: LiteHRNet + post_process: HRNetPostProcess + flip_perm: *flip_perm + num_joints: *num_joints + width: &width 40 + loss: KeyPointMSELoss + use_dark: false + +LiteHRNet: + network_type: wider_naive + freeze_at: -1 + freeze_norm: false + return_idx: [0] + +KeyPointMSELoss: + use_target_weight: true + loss_scale: 1.0 + +#####optimizer +LearningRate: + base_lr: 0.002 + schedulers: + - !PiecewiseDecay + milestones: [170, 200] + gamma: 0.1 + - !LinearWarmup + start_factor: 0.001 + steps: 500 + +OptimizerBuilder: + optimizer: + type: Adam + regularizer: + factor: 0.0 + type: L2 + + +#####data +TrainDataset: + !KeypointTopDownCocoDataset + image_dir: train2017 + anno_path: annotations/person_keypoints_train2017.json + dataset_dir: dataset/coco + num_joints: *num_joints + trainsize: *trainsize + pixel_std: *pixel_std + use_gt_bbox: True + + +EvalDataset: + !KeypointTopDownCocoDataset + image_dir: val2017 + anno_path: annotations/person_keypoints_val2017.json + dataset_dir: dataset/coco + num_joints: *num_joints + trainsize: *trainsize + pixel_std: *pixel_std + use_gt_bbox: True + image_thre: 0.0 + + +TestDataset: + !ImageFolder + anno_path: dataset/coco/keypoint_imagelist.txt + +worker_num: 2 +global_mean: &global_mean [0.485, 0.456, 0.406] +global_std: &global_std [0.229, 0.224, 0.225] +TrainReader: + sample_transforms: + - RandomFlipHalfBodyTransform: + scale: 0.25 + rot: 30 + num_joints_half_body: 8 + prob_half_body: 0.3 + pixel_std: *pixel_std + trainsize: *trainsize + upper_body_ids: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10] + flip_pairs: *flip_perm + - TopDownAffine: + trainsize: *trainsize + - ToHeatmapsTopDown: + hmsize: *hmsize + sigma: 2 + batch_transforms: + - NormalizeImage: + mean: *global_mean + std: *global_std + is_scale: true + - Permute: {} + batch_size: 64 + shuffle: true + drop_last: false + +EvalReader: + sample_transforms: + - TopDownAffine: + trainsize: *trainsize + batch_transforms: + - NormalizeImage: + mean: *global_mean + std: *global_std + is_scale: true + - Permute: {} + batch_size: 16 + +TestReader: + inputs_def: + image_shape: [3, *train_height, *train_width] + sample_transforms: + - Decode: {} + - TopDownEvalAffine: + trainsize: *trainsize + - NormalizeImage: + mean: *global_mean + std: *global_std + is_scale: true + - Permute: {} + batch_size: 1 diff --git a/ppdet/modeling/backbones/lite_hrnet.py b/ppdet/modeling/backbones/lite_hrnet.py index be32c132dc39701aec2ac0a657b39a75c2184f85..fbf1dfe21635b380acb4fbb4c5138028ee0caa0d 100644 --- a/ppdet/modeling/backbones/lite_hrnet.py +++ b/ppdet/modeling/backbones/lite_hrnet.py @@ -271,7 +271,7 @@ class ShuffleUnit(nn.Layer): norm_decay=0.): super(ShuffleUnit, self).__init__() branch_channel = out_channel // 2 - stride = self.stride + self.stride = stride if self.stride == 1: assert ( in_channel == branch_channel * 2, @@ -544,11 +544,11 @@ class LiteHRNetModule(nn.Layer): norm_decay=norm_decay)) return nn.Sequential(*layers) - def _make_naive_branchs(self, - num_branches, - num_blocks, - freeze_norm=False, - norm_decay=0.): + def _make_naive_branches(self, + num_branches, + num_blocks, + freeze_norm=False, + norm_decay=0.): branches = [] for branch_idx in range(num_branches): layers = [] @@ -644,7 +644,7 @@ class LiteHRNetModule(nn.Layer): out = self.layers(x) elif self.module_type == 'NAIVE': for i in range(self.num_branches): - x[i] = self.layers(x[i]) + x[i] = self.layers[i](x[i]) out = x if self.with_fuse: out_fuse = []