search

LikeLy-Journey / SegmenTron

Support PointRend, Fast_SCNN, HRNet, Deeplabv3_plus(xception, resnet, mobilenet), ContextNet, FPENet, DABNet, EdaNet, ENet, Espnetv2, RefineNet, UNet, DANet, HRNet, DFANet, HardNet, LedNet, OCNet, EncNet, DuNet, CGNet, CCNet, BiSeNet, PSPNet, ICNet, FCN, deeplab)
Apache License 2.0
705 stars 162 forks source link

about demo.py #27

Closed y-kl8 closed 4 years ago

y-kl8 commented 4 years ago

somebody know how to use the demo.py?

what is the command?

LikeLy-Journey commented 4 years ago

python tools/demo.py --config-file your-config-file --input-img your-img-path-or-your-img-dir will be ok

andeyeluguo commented 4 years ago

@LikeLy-Journey I found it did not work use your demo. the outputed picture will be fulled with yellow pixels.

WesleyYue commented 4 years ago

@andeyeluguo If you are getting a single class for the entire image you're probably running inference on random weights; make sure you actually loaded a model. The output should print something like "load test model from" https://github.com/LikeLy-Journey/SegmenTron/blob/e6218e6/segmentron/models/model_zoo.py#L34

You need to download weights from the readme and modify the TEST_MODEL_PATH in the config files if you're using the repo out of the box.