xh-liu / hydraplus-net Goto Github PK

Hi, I am interested in you research and would like to use your dataset from this link. However, when I try to unzip data.zip file, it get errors. Can you check if whether the file is corrupted?

Hello,
Could you share your dataset ?
Thanks

Hi!
Kindly provide the evaluation code use in your work to reproduce the result.
Thank.

how run the HydraPlus-Net,thank you

Hi, when I using your prototxt to do some training, I found some layers are not included in my caffe. So could you share your caffe that I can use it to do training? Thanks~

Hi,I'm very interested in your results. Can you share your code for me?

Is it independent when the attention is focused on the 3 branch?

Hello! I love this research! Question: is the PA-100K derived from another pedestrian detection dataset? Or is it collected and generated independently?

Hi Liu,
I tried to duplicate your paper, but I met some questions about the attribute recognition. My net is composed of "res-net" and a "SigmoidCrossEntropyLoss" layer. And the attributes "trousers" and "Age 18-60" occupy a big proportion in "PA-100K" dataset. And it would lead a phenomenon that when I enter a part of a picture the net could also give the labels include "trousers" and "Age 18-60"(I just want a single attribute ). Could you give me some advice to eliminate these distractions? Thank you.

Hi, I need to retrain the model, for an university project, with a custom dataset and using some custom annotations (i.e I will not use the standard classes of PA100K Dataset). How is your annotation file made?

Hi Liu,
I want make sure the construct of AF-Net. Could you help me check it?
I modified the branch behind "ch_concat_3a_chconcat" layer and I just use L = 4.

And this is my prototxt for caffe .

layer {
name: "ch_concat_3a_chconcat"
type: "Concat"
bottom: "conv_3a_1x1"
bottom: "conv_3a_3x3"
bottom: "conv_3a_double_3x3_1"
bottom: "conv_3a_proj"
top: "ch_concat_3a_chconcat"
}

layer {
name: "attention_conv_3b_1x1"
type: "Convolution"
bottom: "ch_concat_3a_chconcat"
top: "attention_conv_3b_1x1"
convolution_param {
num_output: 4
kernel_size: 1
stride: 1
pad: 0
}
}

layer {
name: "slice_attention_conv_3b_1x1"
type: "Slice"
bottom: "attention_conv_3b_1x1"
top: "slice_attention_conv_3b_1x1_0"
top: "slice_attention_conv_3b_1x1_1"
top: "slice_attention_conv_3b_1x1_2"
top: "slice_attention_conv_3b_1x1_3"

slice_param {
axis: 1
slice_point: 1
slice_point: 2
slice_point: 3
slice_point: 4
}
}

layer
{
name: "attention_mul_feature_0"
type: "Eltwise"
bottom: "ch_concat_3a_chconcat"
bottom: "slice_attention_conv_3b_1x1_0"
top: "attention_mul_feature_0"
eltwise_param {
operation: PROD
}
}
layer
{
name: "attention_mul_feature_1"
type: "Eltwise"
bottom: "ch_concat_3a_chconcat"
bottom: "slice_attention_conv_3b_1x1_1"
top: "attention_mul_feature_1"
eltwise_param {
operation: PROD
}
}
layer
{
name: "attention_mul_feature_2"
type: "Eltwise"
bottom: "ch_concat_3a_chconcat"
bottom: "slice_attention_conv_3b_1x1_2"
top: "attention_mul_feature_2"
eltwise_param {
operation: PROD
}
}
layer
{
name: "attention_mul_feature_3"
type: "Eltwise"
bottom: "ch_concat_3a_chconcat"
bottom: "slice_attention_conv_3b_1x1_3"
top: "attention_mul_feature_3"
eltwise_param {
operation: PROD
}
}
layer {
name: "attention_3a_chconcat"
type: "Concat"
bottom: "attention_mul_feature_0"
bottom: "attention_mul_feature_1"
bottom: "attention_mul_feature_2"
bottom: "attention_mul_feature_3"
top: "attention_3a_chconcat"
}
Thank you.

Can you write the readme.txt in detail? You'd better either publish the code or make it more detailed. What do you mean by that? Don't you want someone to run your code?

Hi,
Could you tell me how to create PA100K dataset with my own data?

Hi. I am trying to reproduce the attribute recognition results in Table 2 of your paper. Could you please share your "trainval.prototxt" for your "Main Net"? Thanks.

Hello, I can't find your evaluation code in the attribute recognition, so do you use all 0.5s as thresholds? or do you use specific thresholds for different attributes? Thank you.

Hi,
Can you add a specific license (for the dataset)?
I suggest the CC-BY 4.0 license "Creative Commons — Attribution 4.0 International — CC BY 4.0"

Thanks
Zvi

In the 2nd training stage, we use conv_1x1 generate alpha_i.
Then we use mulitiplication for alpha_i and inception, and get the output.(The output shape: batch x L*C x H x W).
In your paper, you said: Each attentive feature map is then passed through the following blocks thereafter.....
Does it mean use the output feed into the inception module ?
for example, in F(alpha_2), the output
incept_1_alpha_2_output feed into [incept_1, incept_2, incept_3],
incept_2_alpha_2_output feed into [incept_2, incept_3]
and incept_3_alpha_2_output feed into [incept_3]
or all of output feed into [incept_1, incept_2, incept_3]?