yu1206

build.ninja编译，提示torch 中的python_headers.h 文件找不到 头文件#include <Python.h>

1

build.ninja编译，提示torch 中的python_headers.h 文件找不到 头文件#include

missing keys in state_dict

1

Thansk for your work. I use your model. missing keys in state_dict: {'base_model.base.28.bias', 'net_vlad.conv.weight', 'base_model.base.19.bias', 'base_model.base.28.weight', 'base_model.base.24.bias', 'base_model.base.26.weight', 'base_model.base.5.weight', 'base_model.base.7.weight', 'base_model.base.17.bias', 'base_model.base.12.weight', 'net_vlad.centroids', 'base_model.base.7.bias', 'base_model.base.0.bias', 'base_model.base.14.weight', 'base_model.base.5.bias', 'base_model.base.21.bias', 'base_model.base.19.weight',

pitts 250k top-1 88.2%

1

Thansk for your work. I use your model to test data pitts 250k 。 top-1 88.2% top-5 95.4% top-10 96.7% Inconsistent with the paper。

frozen_out/resnet-20707-50-proposals/frozen_inference_graph.pb

1

how are you . Where can I download this file? frozen_out/resnet-20707-50-proposals/model.ckpt

P3 P2 特征一致性训练 是怎么用？

1

流程图看着就是 不同大小的图像进行训练，P3 和P2 特征怎么综合利用？

特征一致性学习 在训练的过程中怎么使用的呢？

label-level consistency learning

1

label-level consistency learning，标签级别的一致性学习 和普通的训练什么区别呢？

On unlabeled data, we first construct a candidate bag for each ground truth g by the traditional IoU-based strategy, where the IoU threshold t is set to a relatively low...

训练时 ，bn层 lr_mult 为什么是0 layer { bottom: "conv1" top: "conv1" name: "bn_conv1" type: "BatchNorm" batch_norm_param { use_global_stats: false } param { lr_mult: 0 } param { lr_mult: 0 } param...