Skip to main content

Table 5 Comparative result with our dataset in Table 3

From: First-person reading activity recognition by deep learning with synthetically generated images

  Baselines Proposed  
  1-NN + FV SVM + FV DCNN  
  Others Reading Others Reading Others Reading  
Scores Precision 19.6 84.0 36.6 84.5 70.8 92.9
for each class Recall 58.7 51.5 47.4 79.5 69.9 95.0
  F-measure 24.6 61.0 33.4 80.5 64.9 93.5
Averages Precision 51.8 60.5 81.9
over classes Recall 55.1 63.5 82.5
  F-measure 42.8 56.9 79.2
  1. Best methods for a given measure are specified in italic type