The Baseline Roadmap

[Kulbear]

• [x] Play with the starter kit: LB 0.892
• [x] Dataset and DataLoader implementation
  • [x] Augmentation
  • [x] Generator implementation of DataLoader
• [x] Adapt SE-ResNet and SE-InceptionV4
• [x] Implement flipping TTA

[Kulbear]

Doodle Classification Note

Training config:

• ~ 75% of 30k / class ( like ~23k / class)
• size 128 x 128
• batch size 128 (can be a bit more larger)
• 500, steps every epoch, so every epoch train with 128 * 500 images
• Adam, lr start at 2e-4, use cyclic LR with cos annealing

  def cos_annealing_lr(initial_lr, cur_epoch, epoch_per_cycle):
  return initial_lr * (np.cos(np.pi * cur_epoch / epoch_per_cycle) + 1) / 2

• 6 cycles, each cycle 25 epochs. By checking the training log I think this schedule should be improved further)

Model used (with best score):

• MobileNet 0.913 (by hflip TTA)
• ResNet18 0.919
• Xception 0.925 (not with the current 3 channel information, I think the current one can give a better result)
• SE-ResNeXt50 (32 x 4d) 0.930

Ideas used:

• imagenet-pretrained weights
• larger images I tried Xception with 224 x 224 image, by using about 20k/class, and just ReduceOnPlaque can get 0.925, I think larger --> better
• grayscale strokes to ordered strokes
• country code information encoding

  c_value = country_mapping.get(df.iloc[i]['countrycode'], 40)
  x[i, :, :, 0] = img + mask * c_value

• highlight first and last stokes (may drop the last and retry)

  if i == 0 or i == len(stroke[0]) - 2:
  color = 255
  _ = cv2.line(img, (stroke[0][i], stroke[1][i]),
               (stroke[0][i + 1], stroke[1][i + 1]), color, lw)

• hflip TTA (good when LB < 0.920, bad when LB >= 0.920)
• last few layers in the model, with dropout rate 0.075 (prob not good)

  def forward(self, x):
      x = self.features(x)
      x = self.relu(x)
      x = self.avg_pool(x)
      x = x.view(x.size(0), -1)
      x = self.dropout(x)
      x = self.last_linear(x)

Possible Models:

Size 1:

• Xception
• ResNet50
• SE ResNeXt50

Size 2:

• SE ResNeXt101
• DPN?
• InceptionResV2

TODO:

• better LR scheduler (this seems to be the most important one)
• update network after several batches to increase the valid batch size
• finetune with different optimizers?
• larger batches
• lstm
  • bi-lstm
  • gru?
  • cnn as feature extractor and rnn accept result from global pooling?
• label denoising

[Kulbear]

[Kulbear]

https://www.kaggle.com/c/human-protein-atlas-image-classification/discussion/74374#437548

train_df_orig=train_df.copy()    
lows = [15,15,15,8,9,10,8,9,10,8,9,10,17,20,24,26,15,27,15,20,24,17,8,15,27,27,27]
for i in lows:
    target = str(i)
    indicies = train_df_orig.loc[train_df_orig['Target'] == target].index
    train_df = pd.concat([train_df,train_df_orig.loc[indicies]], ignore_index=True)
    indicies = train_df_orig.loc[train_df_orig['Target'].str.startswith(target+" ")].index
    train_df = pd.concat([train_df,train_df_orig.loc[indicies]], ignore_index=True)
    indicies = train_df_orig.loc[train_df_orig['Target'].str.endswith(" "+target)].index
    train_df = pd.concat([train_df,train_df_orig.loc[indicies]], ignore_index=True)
    indicies = train_df_orig.loc[train_df_orig['Target'].str.contains(" "+target+" ")].index
    train_df = pd.concat([train_df,train_df_orig.loc[indicies]], ignore_index=True)