search

Owen-Liuyuxuan / visualDet3D

Official Repo for Ground-aware Monocular 3D Object Detection for Autonomous Driving / YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection
https://owen-liuyuxuan.github.io/papers_reading_sharing.github.io/3dDetection/GroundAwareConvultion/
Apache License 2.0
361 stars 76 forks source link

修改输入图像resize的大小 #62

Open monsters-s opened 2 years ago

monsters-s commented 2 years ago

在yolostereo3D中，修改配置文件中的data.rgb_shape,还需要修改配置文件中的其他参数么

Owen-Liuyuxuan commented 2 years ago

If you want to train yolostereo3D with a different image shape:

monsters-s commented 2 years ago

If you want to train yolostereo3D with a different image shape:

  • Change data.rgb_shape.
  • Change anchor sizes anchors.sizes if needed.
  • Change data.augmentation.crop_top if needed.

I also have another question, head_layer.num_features_in=1408, 1408 is a simple hyperparameter or associated with other parameters

Owen-Liuyuxuan commented 2 years ago

The number of input features is associated with the output feature of the neck.