未验证 提交 5298ee05 编写于 作者: W wangxinxin08 提交者: GitHub

fix feat_channel and polish voc config (#2765)

上级 e7623651
......@@ -19,7 +19,6 @@ ResNet:
norm_decay: 0.
PPYOLOFPN:
feat_channels: [512, 512]
drop_block: true
block_size: 3
keep_prob: 0.9
......
......@@ -10,14 +10,6 @@ snapshot_epoch: 83
weights: output/ppyolo_r50vd_dcn_voc/model_final
TrainReader:
batch_transforms:
- BatchRandomResize: {target_size: [320, 352, 384, 416, 448, 480, 512, 544, 576, 608], random_size: True, random_interp: True, keep_ratio: False}
- NormalizeBox: {}
- PadBox: {num_max_boxes: 50}
- BboxXYXY2XYWH: {}
- NormalizeImage: {mean: [0.485, 0.456, 0.406], std: [0.229, 0.224, 0.225], is_scale: True}
- Permute: {}
- Gt2YoloTarget: {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]], downsample_ratios: [32, 16, 8], num_classes: 20}
mixup_epoch: 350
batch_size: 12
......
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