search

microsoft / scene_graph_benchmark

image scene graph generation benchmark
MIT License
382 stars 86 forks source link

predcls ValueError: object labelmap is required, but was not provided #19

Open alice-cool opened 3 years ago

alice-cool commented 3 years ago

hi I want to get some suggestions about the more object label circumstance. I found the project provides us the labelmap file. But the labelmap only provide 50 objects. It is so smaller. I run the coco2014 , the object label 1370 is far bigger than it . So We should only add some id and name to the labelmap.file or we should from scratch to train the object detector of the project ? I have the coco 2014 36 box's label. But I don't know how to get bigger object label labelmap file

If you have way, please help me . Thanks

KeyError: 'broccoli' Killing subprocess 4356

alice-cool commented 3 years ago

image

alice-cool commented 3 years ago

The website didn't have minival annotations 2014

alice-cool commented 3 years ago

ValueError: object labelmap is required, but was not provided

How to get labelmap

alice-cool commented 3 years ago

AttributeError: 'COCODataset' object has no attribute 'get_img_key'

alice-cool commented 3 years ago

image

alice-cool commented 3 years ago

I test 100 images and I got the result. No problem image

MikeDean2367 commented 3 years ago

I test 100 images and I got the result. No problem image

Are you testing on your own image? If so, can you share how you generated test data with your own images? I've tried many things, but it still doesn't work. I would be most grateful if you could answer me.

alice-cool commented 3 years ago

https://github.com/microsoft/scene_graph_benchmark/issues/19#issuecomment-855217314 But because my object detector has 1370 class ,so now I can't get the result. It brokes, because the model neuralmotif only have 150 object label....

alice-cool commented 3 years ago

https://github.com/microsoft/scene_graph_benchmark/issues/19#issuecomment-855217314 I only can give you some idea. The first . you can use demo_tsv.py to generate your file. The second When you run the predcls method. The annotation has no 'relations' attribute. either you leave it or you modify the demo_tsv generate code to make it have the 'relations' attribute.

The third. Because we always no relations attribute . So we should construct false dict like the following. And use it to replace the annotations['relations'] image

If you have more idea about how to use more object label method, can tell me . Now I only make more object label confined to the 150 label....... So it is troublesome Good luck to you https://blog.csdn.net/u012211422/article/details/117733173

dasfinux commented 2 years ago

I test 100 images and I got the result. No problem image

Hello my friend, I wanna generate own dataset, what the labelmap.json look like? and how to generate labelmap.json.

alice-cool commented 2 years ago

I think maybe only use the project provided.  If you make a new labelmap,maybe you should from scratch to train the model such as the number of categories.  labelmap. json can be designated by  test_danfeiX_relation. yaml

the content format is like VGG-SGG-dicts-danfeiX-clipped. json

---原始邮件--- 发件人: @.> 发送时间: 2021年8月11日(星期三) 上午9:37 收件人: @.>; 抄送: @.**@.>; 主题: Re: [microsoft/scene_graph_benchmark] predcls ValueError: object labelmap is required, but was not provided (#19)

I test 100 images and I got the result. No problem

Hello my friend, I wanna generate own dataset, what the labelmap.json look like? and how to generate labelmap.json.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android.