Train GPND on COCO #17

Closed
opened 2019-02-19 14:57:28 +01:00 by 2martens · 1 comment
Owner

Changes to GPND

  • only encoder and decoder
  • activation in last decoder layer is sigmoid
  • padding to same
  • filter sizes changed
  • added pooling layers
  • changed Conv2DTranspose to Conv2D with UpSampling
  • changed kernel size to 3 and stride to 1
  • removed batch normalization layers
  • changed kernel size to 7 and number of filters to 16, 16, 8 and 8, 16, 16 for encoder and decoder
  • added dropout layer with rate 0.25
  • removed dropout layer
  • removed pooling and upsampling layers
  • zsize 64
Changes to GPND - only encoder and decoder - activation in last decoder layer is sigmoid - padding to same - filter sizes changed - added pooling layers - changed Conv2DTranspose to Conv2D with UpSampling - changed kernel size to 3 and stride to 1 - removed batch normalization layers - changed kernel size to 7 and number of filters to 16, 16, 8 and 8, 16, 16 for encoder and decoder - added dropout layer with rate 0.25 - removed dropout layer - removed pooling and upsampling layers - zsize 64
2martens added this to the Trained GPND on SceneNet RGB-D milestone 2019-02-19 14:57:28 +01:00
2martens added the
Resource
label 2019-02-19 14:58:46 +01:00
2martens added the
Uncertainty
label 2019-02-19 15:41:52 +01:00
2martens changed title from Train GPND on SceneNet RGB-D to Train GPND on COCO 2019-04-11 14:44:51 +02:00
Author
Owner

No longer required

No longer required
Sign in to join this conversation.
No Assignees
1 Participants
Notifications
Due Date
The due date is invalid or out of range. Please use the format 'yyyy-mm-dd'.

No due date set.

Dependencies

No dependencies set.

Reference: 2martens/masterthesis#17
No description provided.