未验证 提交 72b7f9ef 编写于 作者: W wangxinxin08 提交者: GitHub

modify transform op of yolo and solo configs (#2260)

上级 1a66a9d2
...@@ -10,17 +10,13 @@ weights: output/fcos_r50_fpn_multiscale_2x_coco/model_final ...@@ -10,17 +10,13 @@ weights: output/fcos_r50_fpn_multiscale_2x_coco/model_final
TrainReader: TrainReader:
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- RandomFlipOp: {prob: 0.5} - RandomFlip: {prob: 0.5}
- NormalizeImageOp: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]} - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
- ResizeImage: - RandomResize: {target_size: [[640, 1333], [672, 1333], [704, 1333], [736, 1333], [768, 1333], [800, 1333]], keep_ratio: true, interp: 1}
target_size: [640, 672, 704, 736, 768, 800] - Permute: {}
max_size: 1333
interp: 1
use_cv2: true
- PermuteOp: {}
batch_transforms: batch_transforms:
- PadBatchOp: {pad_to_stride: 128} - PadBatch: {pad_to_stride: 128}
- Gt2FCOSTarget: - Gt2FCOSTarget:
object_sizes_boundary: [64, 128, 256, 512] object_sizes_boundary: [64, 128, 256, 512]
center_sampling_radius: 1.5 center_sampling_radius: 1.5
......
...@@ -13,24 +13,24 @@ TrainReader: ...@@ -13,24 +13,24 @@ TrainReader:
inputs_def: inputs_def:
num_max_boxes: 90 num_max_boxes: 90
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- MixupOp: {alpha: 1.5, beta: 1.5} - Mixup: {alpha: 1.5, beta: 1.5}
- RandomDistortOp: {} - RandomDistort: {}
- RandomExpandOp: {fill_value: [123.675, 116.28, 103.53]} - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
- RandomCropOp: {} - RandomCrop: {}
- RandomFlipOp: {} - RandomFlip: {}
batch_transforms: batch_transforms:
- BatchRandomResizeOp: - BatchRandomResize:
target_size: [224, 256, 288, 320, 352, 384, 416, 448, 480, 512] target_size: [224, 256, 288, 320, 352, 384, 416, 448, 480, 512]
random_size: True random_size: True
random_interp: True random_interp: True
keep_ratio: False keep_ratio: False
- NormalizeBoxOp: {} - NormalizeBox: {}
- PadBoxOp: {num_max_boxes: 90} - PadBox: {num_max_boxes: 90}
- BboxXYXY2XYWHOp: {} - BboxXYXY2XYWH: {}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
- Gt2YoloTargetOp: - Gt2YoloTarget:
anchor_masks: [[3, 4, 5], [0, 1, 2]] anchor_masks: [[3, 4, 5], [0, 1, 2]]
anchors: [[11, 18], [34, 47], [51, 126], [115, 71], [120, 195], [254, 235]] anchors: [[11, 18], [34, 47], [51, 126], [115, 71], [120, 195], [254, 235]]
downsample_ratios: [32, 16] downsample_ratios: [32, 16]
...@@ -42,10 +42,10 @@ TrainReader: ...@@ -42,10 +42,10 @@ TrainReader:
EvalReader: EvalReader:
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- ResizeOp: {target_size: [320, 320], keep_ratio: False, interp: 2} - Resize: {target_size: [320, 320], keep_ratio: False, interp: 2}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
batch_size: 8 batch_size: 8
drop_empty: false drop_empty: false
...@@ -53,10 +53,10 @@ TestReader: ...@@ -53,10 +53,10 @@ TestReader:
inputs_def: inputs_def:
image_shape: [3, 320, 320] image_shape: [3, 320, 320]
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- ResizeOp: {target_size: [320, 320], keep_ratio: False, interp: 2} - Resize: {target_size: [320, 320], keep_ratio: False, interp: 2}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
batch_size: 1 batch_size: 1
epoch: 270 epoch: 270
......
...@@ -13,24 +13,24 @@ TrainReader: ...@@ -13,24 +13,24 @@ TrainReader:
inputs_def: inputs_def:
num_max_boxes: 90 num_max_boxes: 90
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- MixupOp: {alpha: 1.5, beta: 1.5} - Mixup: {alpha: 1.5, beta: 1.5}
- RandomDistortOp: {} - RandomDistort: {}
- RandomExpandOp: {fill_value: [123.675, 116.28, 103.53]} - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
- RandomCropOp: {} - RandomCrop: {}
- RandomFlipOp: {} - RandomFlip: {}
batch_transforms: batch_transforms:
- BatchRandomResizeOp: - BatchRandomResize:
target_size: [224, 256, 288, 320, 352, 384, 416, 448, 480, 512] target_size: [224, 256, 288, 320, 352, 384, 416, 448, 480, 512]
random_size: True random_size: True
random_interp: True random_interp: True
keep_ratio: False keep_ratio: False
- NormalizeBoxOp: {} - NormalizeBox: {}
- PadBoxOp: {num_max_boxes: 90} - PadBox: {num_max_boxes: 90}
- BboxXYXY2XYWHOp: {} - BboxXYXY2XYWH: {}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
- Gt2YoloTargetOp: - Gt2YoloTarget:
anchor_masks: [[3, 4, 5], [0, 1, 2]] anchor_masks: [[3, 4, 5], [0, 1, 2]]
anchors: [[11, 18], [34, 47], [51, 126], [115, 71], [120, 195], [254, 235]] anchors: [[11, 18], [34, 47], [51, 126], [115, 71], [120, 195], [254, 235]]
downsample_ratios: [32, 16] downsample_ratios: [32, 16]
...@@ -42,10 +42,10 @@ TrainReader: ...@@ -42,10 +42,10 @@ TrainReader:
EvalReader: EvalReader:
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- ResizeOp: {target_size: [320, 320], keep_ratio: False, interp: 2} - Resize: {target_size: [320, 320], keep_ratio: False, interp: 2}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
batch_size: 8 batch_size: 8
drop_empty: false drop_empty: false
...@@ -53,10 +53,10 @@ TestReader: ...@@ -53,10 +53,10 @@ TestReader:
inputs_def: inputs_def:
image_shape: [3, 320, 320] image_shape: [3, 320, 320]
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- ResizeOp: {target_size: [320, 320], keep_ratio: False, interp: 2} - Resize: {target_size: [320, 320], keep_ratio: False, interp: 2}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
batch_size: 1 batch_size: 1
epoch: 270 epoch: 270
......
...@@ -11,27 +11,27 @@ weights: output/ppyolo_r18vd_coco/model_final ...@@ -11,27 +11,27 @@ weights: output/ppyolo_r18vd_coco/model_final
TrainReader: TrainReader:
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- MixupOp: {alpha: 1.5, beta: 1.5} - Mixup: {alpha: 1.5, beta: 1.5}
- RandomDistortOp: {} - RandomDistort: {}
- RandomExpandOp: {fill_value: [123.675, 116.28, 103.53]} - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
- RandomCropOp: {} - RandomCrop: {}
- RandomFlipOp: {} - RandomFlip: {}
batch_transforms: batch_transforms:
- BatchRandomResizeOp: - BatchRandomResize:
target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608] target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
random_size: True random_size: True
random_interp: True random_interp: True
keep_ratio: False keep_ratio: False
- NormalizeBoxOp: {} - NormalizeBox: {}
- PadBoxOp: {num_max_boxes: 50} - PadBox: {num_max_boxes: 50}
- BboxXYXY2XYWHOp: {} - BboxXYXY2XYWH: {}
- NormalizeImageOp: - NormalizeImage:
mean: [0.485, 0.456, 0.406] mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225] std: [0.229, 0.224, 0.225]
is_scale: True is_scale: True
- PermuteOp: {} - Permute: {}
- Gt2YoloTargetOp: - Gt2YoloTarget:
anchor_masks: [[3, 4, 5], [0, 1, 2]] anchor_masks: [[3, 4, 5], [0, 1, 2]]
anchors: [[10, 14], [23, 27], [37, 58], [81, 82], [135, 169], [344, 319]] anchors: [[10, 14], [23, 27], [37, 58], [81, 82], [135, 169], [344, 319]]
downsample_ratios: [32, 16] downsample_ratios: [32, 16]
...@@ -42,10 +42,10 @@ TrainReader: ...@@ -42,10 +42,10 @@ TrainReader:
EvalReader: EvalReader:
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- ResizeOp: {target_size: [512, 512], keep_ratio: False, interp: 2} - Resize: {target_size: [512, 512], keep_ratio: False, interp: 2}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
batch_size: 8 batch_size: 8
drop_empty: false drop_empty: false
...@@ -53,10 +53,10 @@ TestReader: ...@@ -53,10 +53,10 @@ TestReader:
inputs_def: inputs_def:
image_shape: [3, 512, 512] image_shape: [3, 512, 512]
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- ResizeOp: {target_size: [512, 512], keep_ratio: False, interp: 2} - Resize: {target_size: [512, 512], keep_ratio: False, interp: 2}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
batch_size: 1 batch_size: 1
epoch: 270 epoch: 270
......
...@@ -20,17 +20,17 @@ LearningRate: ...@@ -20,17 +20,17 @@ LearningRate:
TrainReader: TrainReader:
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- Poly2Mask: {} - Poly2Mask: {}
- RandomResizeOp: {interp: 1, - RandomResize: {interp: 1,
target_size: [[640, 1333], [672, 1333], [704, 1333], [736, 1333], [768, 1333], [800, 1333]], target_size: [[640, 1333], [672, 1333], [704, 1333], [736, 1333], [768, 1333], [800, 1333]],
keep_ratio: True} keep_ratio: True}
- RandomFlipOp: {} - RandomFlip: {}
- NormalizeImageOp: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]} - NormalizeImage: {is_scale: true, mean: [0.485,0.456,0.406], std: [0.229, 0.224,0.225]}
- PermuteOp: {} - Permute: {}
batch_transforms: batch_transforms:
- PadBatchOp: {pad_to_stride: 32} - PadBatch: {pad_to_stride: 32}
- Gt2Solov2TargetOp: {num_grids: [40, 36, 24, 16, 12], - Gt2Solov2Target: {num_grids: [40, 36, 24, 16, 12],
scale_ranges: [[1, 96], [48, 192], [96, 384], [192, 768], [384, 2048]], scale_ranges: [[1, 96], [48, 192], [96, 384], [192, 768], [384, 2048]],
coord_sigma: 0.2} coord_sigma: 0.2}
batch_size: 2 batch_size: 2
......
...@@ -13,24 +13,24 @@ TrainReader: ...@@ -13,24 +13,24 @@ TrainReader:
inputs_def: inputs_def:
num_max_boxes: 50 num_max_boxes: 50
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- MixupOp: {alpha: 1.5, beta: 1.5} - Mixup: {alpha: 1.5, beta: 1.5}
- RandomDistortOp: {} - RandomDistort: {}
- RandomExpandOp: {fill_value: [123.675, 116.28, 103.53]} - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
- RandomCropOp: {} - RandomCrop: {}
- RandomFlipOp: {} - RandomFlip: {}
batch_transforms: batch_transforms:
- BatchRandomResizeOp: - BatchRandomResize:
target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608] target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
random_size: True random_size: True
random_interp: True random_interp: True
keep_ratio: False keep_ratio: False
- NormalizeBoxOp: {} - NormalizeBox: {}
- PadBoxOp: {num_max_boxes: 50} - PadBox: {num_max_boxes: 50}
- BboxXYXY2XYWHOp: {} - BboxXYXY2XYWH: {}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
- Gt2YoloTargetOp: - Gt2YoloTarget:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]] anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], [59, 119], [116, 90], [156, 198], [373, 326]] anchors: [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], [59, 119], [116, 90], [156, 198], [373, 326]]
downsample_ratios: [32, 16, 8] downsample_ratios: [32, 16, 8]
......
...@@ -19,24 +19,24 @@ TrainReader: ...@@ -19,24 +19,24 @@ TrainReader:
inputs_def: inputs_def:
num_max_boxes: 50 num_max_boxes: 50
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- MixupOp: {alpha: 1.5, beta: 1.5} - Mixup: {alpha: 1.5, beta: 1.5}
- RandomDistortOp: {} - RandomDistort: {}
- RandomExpandOp: {fill_value: [123.675, 116.28, 103.53]} - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
- RandomCropOp: {} - RandomCrop: {}
- RandomFlipOp: {} - RandomFlip: {}
batch_transforms: batch_transforms:
- BatchRandomResizeOp: - BatchRandomResize:
target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608] target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
random_size: True random_size: True
random_interp: True random_interp: True
keep_ratio: False keep_ratio: False
- NormalizeBoxOp: {} - NormalizeBox: {}
- PadBoxOp: {num_max_boxes: 50} - PadBox: {num_max_boxes: 50}
- BboxXYXY2XYWHOp: {} - BboxXYXY2XYWH: {}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
- Gt2YoloTargetOp: - Gt2YoloTarget:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]] anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], [59, 119], [116, 90], [156, 198], [373, 326]] anchors: [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], [59, 119], [116, 90], [156, 198], [373, 326]]
downsample_ratios: [32, 16, 8] downsample_ratios: [32, 16, 8]
......
...@@ -13,24 +13,24 @@ TrainReader: ...@@ -13,24 +13,24 @@ TrainReader:
inputs_def: inputs_def:
num_max_boxes: 50 num_max_boxes: 50
sample_transforms: sample_transforms:
- DecodeOp: {} - Decode: {}
- MixupOp: {alpha: 1.5, beta: 1.5} - Mixup: {alpha: 1.5, beta: 1.5}
- RandomDistortOp: {} - RandomDistort: {}
- RandomExpandOp: {fill_value: [123.675, 116.28, 103.53]} - RandomExpand: {fill_value: [123.675, 116.28, 103.53]}
- RandomCropOp: {} - RandomCrop: {}
- RandomFlipOp: {} - RandomFlip: {}
batch_transforms: batch_transforms:
- BatchRandomResizeOp: - BatchRandomResize:
target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608] target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608]
random_size: True random_size: True
random_interp: True random_interp: True
keep_ratio: False keep_ratio: False
- NormalizeBoxOp: {} - NormalizeBox: {}
- PadBoxOp: {num_max_boxes: 50} - PadBox: {num_max_boxes: 50}
- BboxXYXY2XYWHOp: {} - BboxXYXY2XYWH: {}
- NormalizeImageOp: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True} - NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- PermuteOp: {} - Permute: {}
- Gt2YoloTargetOp: - Gt2YoloTarget:
anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]] anchor_masks: [[6, 7, 8], [3, 4, 5], [0, 1, 2]]
anchors: [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], [59, 119], [116, 90], [156, 198], [373, 326]] anchors: [[10, 13], [16, 30], [33, 23], [30, 61], [62, 45], [59, 119], [116, 90], [156, 198], [373, 326]]
downsample_ratios: [32, 16, 8] downsample_ratios: [32, 16, 8]
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册