Please enable JavaScript.
Coggle requires JavaScript to display documents.
Carreira and Zisserman (2017), Quo Vadis, Action Recognition? A New Model…
Carreira and Zisserman (2017), Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset
Kinetics Dataset
-
-
Used for pre-training models and testing effectiveness of transfer learning for other benchmark datasets.
ConvNet + LSTM
Approach based on using image features from state-of-the-art image classification models like inception and then using Recurrent Neural Networks like LSTM to model temporal information
-
-
-
-
-
Model Evaluation
-
-
Two-Stream Inflated 3D ConvNets proved to be the best performing model across all of the test scenarios
-
-