Open zxzhijia opened 4 years ago
Here is another 3D CNN:
This SeuTao, who is the current winner, had posted a short notebook with his github link below using 3DConvNet 9 days ago. @yanzhang9202 We might try this later :)
https://www.kaggle.com/c/trends-assessment-prediction/discussion/147797
https://github.com/SeuTao/RSNA2019_Intracranial-Hemorrhage-Detection/tree/master/3DNet
Wenjing has asked a good question :)
@yanzhang9202 It seems from Wenjing's understanding, SeuTao is using 3DConvNet + FNC + loading features together that gives the SOTA so far. What do you think?
The paper below uses a 3D ResNet CNN. But not sure what's the score.
https://www.kaggle.com/rftexas/gpu-parallel-training-3d-resnet-cnn