| |
|
|
| |
| split = 'variant_test' ; |
| |
| |
| |
|
|
| switch 1 |
| case 1 |
| |
| images = {'0900914'} ; |
| labels = {'747-400'} ; |
| scores = 1 ; |
| case 2 |
| |
| |
| [images, labels] = textread(fullfile('data', ['images_' split '.txt']), '%7s%*1s%s', 'delimiter', '\n', 'whitespace', '') ; |
| scores = ones(size(labels)) ; |
| case 3 |
| |
| |
| |
| [images0, labels0] = textread(fullfile('data', ['images_' split '.txt']), '%7s%*1s%s', 'delimiter', '\n', 'whitespace', '') ; |
| n = numel(images0) ; |
| clear images labels scores ; |
| for ci = 1:100 |
| images{ci} = 1:n ; |
| labels{ci} = repmat(ci,1,n) ; |
| scores{ci} = randn(1,n) ; |
| end |
| images = [images{:}] ; |
| labels = [labels{:}] ; |
| scores = [scores{:}] ; |
| end |
|
|
| [confusion, results] = evaluation('data', split, images, labels, scores) ; |
|
|
| figure(1) ; clf ; |
| imagesc(confusion) ; axis tight equal ; |
| xlabel('predicted') ; |
| ylabel('ground truth') ; |
| title(sprintf('mean accuracy: %.2f %%\n', mean(diag(confusion))*100)) ; |
|
|