KO
|
EN
gitlite — search
Search
#javascript
#python
#hacktoberfest
#react
#ai
#typescript
#llm
#go
#golang
#android
#machine-learning
#rust
#deep-learning
#linux
centernet-pasvoc
★ 9
Open GitHub ↗
Centernet Pytorch PasVOC
Download README (.md)
Explore Similar Repositories
svelte-markup-walker
:
A Svelte preprocessor that allows to walk through the AST generated by the svelte parser and modify the markup with MagicString.
ShadowsocksR-win-4.9.0
:
ShadowsocksR-win-4.9.0
crud-php7
:
CRUD PHP 7 MySQLi
zabbix_telegram_webhook
:
Send message from Zabbix 4.4 to Telegram via Webhook
silvesterwali
:
FullStack Developer working with laravel, nuxtjs, vuejs, expressjs, nodejs,mongodb,SQL
// repository documentation
Was this content helpful?
★ 0
(0 ratings)
Select Rating:
★
★
★
★
★
Submit Feedback
Recent Feedback
×
Download README
Do you want to download the
README.md
file for
centernet-pasvoc
?
Download (.md)
# one-stage-object-detection one stage object detection include yolov1 yolov2 yolov3 object as point reference https://github.com/xingyizhou/CenterNet ## the simple train method which not contain DCN * modify the code in the models __init__.py ```python # from .dlav0 import get_pose_net #from .resnet_dcn import get_pose_net # from .pose_dla_dcn import get_pose_net from .msra_resnet import get_pose_net __all__=( 'get_pose_net', ``` ## the code need to complile the DCNv2,this code is compliled in windows 10,VS2015 python3.6 ## the file directory is: ## detect.py test the original model(train in coco dataset) * testvoc.py test the result of train using the PASVOC dataset ## data preprocess * download PASVOC dataset in the data dir move the voc_label.py to the data dir ``` python voc_label.py ``` could generate the *.txt like 2007_test.txt 2007_train.txt 2007_val.txt 2012_train.txt 2012_val.txt ```sh cat 2007_train.txt 2012_train.txt 2012_val.txt 2007_val.txt>train.txt ``` ``` cat 2007_test.txt >val.txt ``` ## train train code is in the experiments ```sh sh pasvoc_384_origin.sh ``` using the resnet18dcn the mAP is 0.69 using the resnet50dcn the mAP is 0.73