Hello, antran. I've read both your paper and code. But I cannot find the code implementation of calculating actionness, such as "sudden change, temporal synchrony, repetitive motion and image saliency". except function "computeActionnessWeightedChi2Kernel" etc preparing for svm. I seems to find that in
- /BoW_frameworks/Saliency_pool_BoW_newAPI/tbox2/loadRegionIndex.m:
- 13 % change this command according to the dataset
- 14 S = load(fullfile(params.regionindex, action,...
- 15: strcat(Actioness2DenseFea_, data_file_name(1:end-4), '_K', num2str(params.num_regions))));
- 16 regionIndex = S.RegionIndex;
- 17 end
and
- /BoW_frameworks/Saliency_pool_BoW_newAPI/tbox2/computeActionnessWeights.m:
- 17 for i = 1:length(all_train_files)
- 18 load(fullfile(params.regionindex, params.actions{all_train_labels(i)}, ...
- 19: strcat(ActionessOfRegions_, all_train_files{i}(1:end-4), '_K', num2str(params.num_regions))));
- 20 weights = weights + ActionnessValue(1:num_grids);
- 21 end
Are Actioness2DenseFea_ and ActionessOfRegions_ produced by your another codes not included in this project?
Thank you.
Hello, antran. I've read both your paper and code. But I cannot find the code implementation of calculating actionness, such as "sudden change, temporal synchrony, repetitive motion and image saliency". except function "computeActionnessWeightedChi2Kernel" etc preparing for svm. I seems to find that in
and
Are Actioness2DenseFea_ and ActionessOfRegions_ produced by your another codes not included in this project?
Thank you.