profile
viewpoint

MiaoZhang0525/NSAS_FOR_CVPR 8

NSAS code for CVPR review

MiaoZhang0525/ENNAS_MASTER 2

Codes for IJCAI2020 paper "One-Shot Neural Architecture Search via Novelty Driven Sampling"

MiaoZhang0525/HDBO-SDR 2

Code for "High Dimensional Bayesian Optimization via Supervised Dimension Reduction"

MiaoZhang0525/anonymous_github 1

Anonymous Github is a proxy server to support anonymous browsing of Github repositories for open-science code and data.

MiaoZhang0525/Awesome-NAS 1

A curated list of neural architecture search (NAS) resources.

MiaoZhang0525/A-Quality-Diversity-based-Genetic-Algorithm-with-Redundancy-Measurement-for-Feature-Selection 0

the code for our paper "A Quality-Diversity based Genetic Algorithm with Redundancy Measurement for Feature Selection"

MiaoZhang0525/A-Reference-Direction-and-Entropy-based-Evolutionary-Algorithm-for-Many-Objective-Optimization 0

the code for our paper "A Reference Direction and Entropy based Evolutionary Algorithm for Many-Objective Optimization"

MiaoZhang0525/agem 0

Official implementation of the Averaged Gradient Episodic Memory (A-GEM) in Tensorflow

MiaoZhang0525/cgp-cnn 0

A Genetic Programming Approach to Designing CNN Architectures, In GECCO 2017 (oral presentation, Best Paper Award)

push eventMiaoZhang0525/EENAS_for_NeurIPS2020

MiaoZhang0525

commit sha b06601e0e6675da2d5870fe083c301f4ffb7d048

Create DVAE-MODEL-600.png

view details

push time in 7 days

push eventMiaoZhang0525/EENAS_for_NeurIPS2020

MiaoZhang0525

commit sha bb875985e859b520736e50fc682b5a00baf4e545

Add darts search space results

view details

push time in 7 days

create barnchMiaoZhang0525/EENAS_for_NeurIPS2020

branch : master

created branch time in 20 days

created repositoryMiaoZhang0525/EENAS_for_NeurIPS2020

created time in 20 days

push eventMiaoZhang0525/MiaoZhang.github.io

MiaoZhang0525

commit sha cdc73d6a94c64bb39993b61470732a63a7830aff

Update index.md

view details

push time in a month

create barnchMiaoZhang0525/MiaoZhang.github.io

branch : gh-pages

created branch time in a month

create barnchMiaoZhang0525/MiaoZhang.github.io

branch : master

created branch time in a month

created repositoryMiaoZhang0525/MiaoZhang.github.io

created time in a month

created repositoryMiaoZhang0525/MiaoZhang

created time in a month

push eventMiaoZhang0525/NSAS_FOR_CVPR

MiaoZhang0525

commit sha 362b6299ae18a5a74a58f9948d1d094d734403ca

rename

view details

push time in a month

issue openedprincewang1994/RFCN_CoupleNet.pytorch

Is that possible to share you trained model on COCO?

Hi, really appreciate that you share your works. That beneficiate us a lot. I currently using your code to solve our own datasets. I want to use the trained mode on COCO as the initial state. Is it possible to share your trained model on COCO.

That will save a lot of time for us.

Sincerely.

created time in 2 months

issue commentsoo89/CSD-RFCN

Request for the trained model on COCO or VOC

I removed the trained model on my desktop.

I have a link for NeurIPS supplementary and you can get some trained model.

https://drive.google.com/drive/folders/1T6UrU4oU4gL9FCDAauCsHB0Z0gbcAHHp?usp=sharing

Thanks for your help. That is so nice of you. I got these models which are trained on labeled VOC and unlabeled COCO. Is that possible for you to share RFCN trained with labeled COCO? Or give some suggestions?

That will be very helpful for me, and save my time.

Sincerely.

MiaoZhang0525

comment created time in 2 months

issue openedsoo89/CSD-RFCN

Request for the trained model on COCO or VOC

Hi,

Really thanks for your share these wonderful codes, and they really help me a lot. Currently, I am also working on this.

However, if you could provide a pretrained model (especially on COCO), it will greatly help us, and save our time.

Is it possible to share a link of pretrained model on COCO? based RFCN is also OK.

Many many thanks.

Sincerely

created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

I used 4 GPUs and 4 mini-batch. In my case, there is only one value in semi_check.

I think I found the reason. Many thanks! Sincerely!

MiaoZhang0525

comment created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

All samples are applied to horizontally flip. Please check our paper in detail.

I got it. Really thanks for your reply. That is very helpful to me.

I found that your use a function int(semi_check) in https://github.com/soo89/CSD-RFCN/blob/45497e9b69ac6b0d962e1efbe6150ed8ca4188b0/lib/model/rfcn/rfcn_consistency.py#L111

However, the semicheck is an array-like ([1,0,0,0]) in your code, and using int (semi_check) will get an error. How to fix it? Can I simply use int (semi_check[0])??

Many thanks, Sincerely.

MiaoZhang0525

comment created time in 2 months

issue openedRebornL/RFCN-pytorch.1.0

Extremely low mAP

Hi, I implement your code, but my mAP is very very low. can you explain the mAP during the training epochs?

Many thanks!!

Sincerely.

created time in 2 months

issue commentRebornL/RFCN-pytorch.1.0

can you share the mAP ?

hello, I am also implement RFCN based Jianwei Yang's fasterrcnn, but my mAP is low. can you share the mAP?

Hi, have you solved this problem? My mAP is also very low and trained very slow.

HZNUJeffreyRen

comment created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

What I want to say that you need to check whether it is well flipped (original and flipped image) and outputs are different values.

OK! I got it! Really appreciate! One last simple and stupid question, so how do you determine which image's ['flipped'] should be true? Is it random? Sorry to bother you. Many many thanks.

MiaoZhang0525

comment created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

So you mean that I need to change the clone() in rfcn_consistency.py to deepcopy?

Really thanks!

MiaoZhang0525

comment created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

Wher

Could you check output values and whether the second image batch is flipped?

I think that you missed the deepcopy.

Where do I need to use the deepcopy? I did not found that you used this function in the train_csd.

Many many thanks!!

MiaoZhang0525

comment created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

Could you check output values and whether the second image batch is flipped?

I think that you missed the deepcopy.

So in the training, the ['flipped'] for the labeled image and unlabeled image should be True?

MiaoZhang0525

comment created time in 2 months

issue commentsoo89/CSD-RFCN

why the consistency_loss is always 0??

I think the problem is that the line 111, int(semi_check) in the rfcn_consistency.py

semi_check=[1,0,0,0]. So int(semi_check) should not work

MiaoZhang0525

comment created time in 2 months

issue openedsoo89/CSD-RFCN

why the consistency_loss is always 0??

Hi, really appreciate for providing your codes.

However, when I reproduce your code., I just found the consistency loss is always zero. Do you know why this happens?

Many thanks.

created time in 2 months

created repositoryMiaoZhang0525/haogeshige

bbbb

created time in 2 months

push eventMiaoZhang0525/NSAS_FOR_CVPR

MiaoZhang0525

commit sha e420ca1ec3836e20c36e96282be0d81f2786e749

add more random seed

view details

push time in 2 months

issue openedsoo89/CSD-RFCN

load unlabeled images

In line 239 in train_csd: "unlabel_imdb, unlabel_roidb, unlabel_ratio_list, unlabel_ratio_index = combined_roidb(args.imdb_name_unlabel)"

You will load the label information into the unlabel_imdb and unlabel_roidb. It should not load the labeled information, isn't it??

I am very curious that why you load these label information of pascal-voc 2012 dataset.

Hope to get your early response.

Many thanks.

created time in 2 months

push eventMiaoZhang0525/NSAS_FOR_CVPR

MiaoZhang0525

commit sha 5e6b6a5ae81ae7cb97861362553c8ec6c8fdb2a6

add PDARTS setting based imagenet results

view details

push time in 2 months

issue commentShadowXZT/pytorch_RFCN

KeyError: 'chair'

Hi, 朋友你好呀!!你解决这个问题了吗? 我试着网上的方法,结果返回空集了。你怎么弄的啊

C2H5OHlife

comment created time in 3 months

issue commentjwyang/faster-rcnn.pytorch

pytorch 1.0 version - error on custom COCO dataset

Using custom coco dataset, errors happend

after filtering, there are 0 images...
0 roidb entries
Loading pretrained weights from data/pretrained_model/resnet101_caffe.pth
Traceback (most recent call last):
  File "trainval_net.py", line 373, in <module>
    save_name = os.path.join(output_dir, 'faster_rcnn_{}_{}_{}.pth'.format(args.session, epoch, step))
NameError: name 'step' is not defined

could you drop your custom data here code because you have to modify

Have you guys solved it??

EMCP

comment created time in 3 months

more