inductive visual localisation factorised training for
play

Inductive Visual Localisation: Factorised Training for Superior - PowerPoint PPT Presentation

Inductive Visual Localisation: Factorised Training for Superior Generalisation Ankush Gupta Andrea Vedaldi Andrew Zisserman Visual Geometry Group (VGG) University of Oxford 1 BMVC 2018, Newcastle upon Tyne | Ankush Gupta RNNs have a


  1. Inductive Visual Localisation: Factorised Training for Superior Generalisation Ankush Gupta Andrea Vedaldi Andrew Zisserman Visual Geometry Group (VGG) University of Oxford 1 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  2. RNNs have a problem. Poor generalization to sequence lengths beyond those in the training set. Testing Training 2 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  3. Example: Enumerative Counting Counting objects one-by-one. Stop? 0 0 0 1 Training Total count = 3 3 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  4. Example: Enumerative Counting Failure when tested on >3 length input Stop? 0 0 0 0 1 Testing Total count = 6 4 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  5. Why? Non-interpretable recurrent state (s t ) which is trained end-to-end may not learn the correct loop-invariant. 5 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  6. Our Solution 1. Train for one-step inductive updates (not end-to-end). 2. Restrict the recurrent state to a spatial-memory map, which tracks the progress made so far. 6 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  7. Inductive Training Train for one-step end-to-end updates Stop? input image Spatial memory Updated map memory 7 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  8. Results: Enumerative Counting Coloured Shapes & DOTA Airplanes train on 3-5 objects , test on >5 objects 8 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  9. Multi-line Text Recognition Read one line at each step 9 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  10. Results: Multi-line Text Recognition Synth Text Blocks train on 1-4 lines , test on up to 10 lines 10 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  11. Results: Multi-line Text Recognition Vs. State-of-the-art @ ICDAR 2013 Blocks outperform (in terms of Recall, F-score) 11 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

  12. Inductive Visual Localisation: Factorised Training for Superior Generalisation Ankush Gupta Andrea Vedaldi Andrew Zisserman Visual Geometry Group (VGG) University of Oxford #111 Poster 12 BMVC 2018, Newcastle upon Tyne | Ankush Gupta

Recommend


More recommend