Supplementary material
Action Recognition based on Efficient Deep Feature Learning in the Spatio-Temporal Domain
Introduction
Results
Code
Introduction
On this page, we provide some additional results for the UCF-101 and hmdb-51 dataset.
Results
We show some results for the third test split of the UCF-101 and HMDB-51 dataset. Further results for UCF101 can be seen by clicking here and for HMDB-51 here.
Qualitative results
for the UCF-101 dataset
The figures below contain 10 frames uniformly spaced in time for each video. First
row (green color) shows the ground-truth followed by predictions in
decreasing level of confidence. Blue and red show correct and incorrect
predictions, respectively.
Qualitative results
for the HMDB-51 dataset
The figures below contain 10 frames uniformly spaced in time for each video. First
row (green color) shows the ground-truth followed by predictions in
decreasing level of confidence. Blue and red show correct and incorrect
predictions, respectively.
Code
For implementation details please contact the corresponding author: Farzad Husain (shusain@iri.upc.edu)