Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on May 18, 2018

• feat: Add TF flags for config and name some tensors

  

  gdangelo

  committed
  da42850

  Copy full SHA for da42850

• feat: Linearly scales image to have zero mean and unit norm

  

  gdangelo

  committed
  4326512

  Copy full SHA for 4326512

Commits on May 13, 2018

• feat: Set VGG16 as U-Net contracting path

  

  gdangelo

  committed
  e75d919

  Copy full SHA for e75d919

• feat: Build UNET architecture

  

  gdangelo

  committed
  1b18fc8

  Copy full SHA for 1b18fc8

Commits on May 12, 2018

• feat: Visualize scalar summaries with Tensorboard

  

  gdangelo

  committed
  22b5156

  Copy full SHA for 22b5156

• feat: Visualize the Graph with Tensorboard

  

  gdangelo

  committed
  f58b6e1

  Copy full SHA for f58b6e1

Commits on May 11, 2018

• feat: Do not clean training directory by default

  

  gdangelo

  committed
  c0c7b79

  Copy full SHA for c0c7b79

• feat: Add L2 regularization for training

  

  gdangelo

  committed
  2c8655f

  Copy full SHA for 2c8655f

Commits on May 4, 2018

• chore: Move NN scripts in models/ folder

  

  gdangelo

  committed
  059b00d

  Copy full SHA for 059b00d

Commits on May 3, 2018

• refact: Split code in distinct functions

  

  gdangelo

  committed
  84bc518

  Copy full SHA for 84bc518

• feat: Downscale images before training to 256x256

  

  gdangelo

  committed
  29e111b

  Copy full SHA for 29e111b

Commits on May 2, 2018

• fix: Add train.py that has been accidentally removed

  

  gdangelo

  committed
  4d05ab9

  Copy full SHA for 4d05ab9

• refact: Build training TF graph and generate inputs like the TF cifar10 example

  

  gdangelo

  committed
  c74ab38

  Copy full SHA for c74ab38

• feat: Return proba from FCN model and fix backward conv params

  

  gdangelo

  committed
  c1666f4

  Copy full SHA for c1666f4

• chore: Remove unused example images

  

  gdangelo

  committed
  5a86c84

  Copy full SHA for 5a86c84

• docs: Fix a typo in comments

  

  gdangelo

  committed
  7f0d412

  Copy full SHA for 7f0d412

Commits on Apr 20, 2018

• refact: Define run directory from flags

  

  gdangelo

  committed
  565fdef

  Copy full SHA for 565fdef

• feat: Define model to use for training from flags

  

  gdangelo

  committed
  8e3bdae

  Copy full SHA for 8e3bdae

• feat: Add instance ID in TF tensor labels

  

  gdangelo

  committed
  50ff50d

  Copy full SHA for 50ff50d

• refact: Move utils/helpers functions in a different file

  

  gdangelo

  committed
  97659d5

  Copy full SHA for 97659d5

• docs: Add comments in main file

  

  gdangelo

  committed
  6b970ed

  Copy full SHA for 6b970ed

• feat: Load image by padding them to respect network requirements

  

  gdangelo

  committed
  6d42383

  Copy full SHA for 6d42383

Commits on Apr 18, 2018

• feat: Implement training pipeline for FCN

  

  gdangelo

  committed
  80b3b98

  Copy full SHA for 80b3b98

• feat: Build FCN32, FCN16, and FCN8 based on VGG16

  

  gdangelo

  committed
  854a073

  Copy full SHA for 854a073

Commits on Apr 13, 2018

• feat: Add TF training operation (loss + SGD optimizer)

  

  gdangelo

  committed
  ac25034

  Copy full SHA for ac25034

• feat: Implement generator to create dataset batches

  

  gdangelo

  committed
  e2084b8

  Copy full SHA for e2084b8

• feat: Add settings flags for building and training network

  

  gdangelo

  committed
  3114066

  Copy full SHA for 3114066

• feat: Return logits for FCN-32 and add num_classes as input argument of fcn_vgg_16

  

  gdangelo

  committed
  060e46a

  Copy full SHA for 060e46a

• feat: Return only layers up to the last pooling layer

  

  gdangelo

  committed
  5b97873

  Copy full SHA for 5b97873

Commits on Apr 12, 2018

• feat: Build FCN with VGG16 architecture

  

  gdangelo

  committed
  a2d44e1

  Copy full SHA for a2d44e1

• chore: Update .gitignore

  

  gdangelo

  committed
  30c292b

  Copy full SHA for 30c292b

• feat: Test VGG16 and VGG19 implementation

  

  gdangelo

  committed
  4ee7773

  Copy full SHA for 4ee7773

• feat: Load pretrained weights from VGG16 and VGG19

  

  gdangelo

  committed
  85ba583

  Copy full SHA for 85ba583

• feat: Add VGG19 model to class

  

  gdangelo

  committed
  88b6668

  Copy full SHA for 88b6668

• refact: Implement VGG models in a class

  

  gdangelo

  committed
  e4fbb4d

  Copy full SHA for e4fbb4d

Pagination

Previous

Next