Repository navigation
Expand file tree
/
Copy pathREADME.txt
More file actions
23 lines (18 loc) · 1.35 KB
/
Copy pathREADME.txt
File metadata and controls
23 lines (18 loc) · 1.35 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
The script demo.m offers an example of the results obtained by the SR-Clustering algorithm.
Before running the demo modify consequently the following parameters:
loadParametersDemo.m
CNN_params.caffe_path: absolute path to the '/path_to_caffe_installation/matlab/caffe' directory
CNN_params.use_gpu: should only be '1' if your Caffe installation is enabled for GPU computation
CNN_params.model_file: absolute path to CaffeNet model (must be downloaded from Caffe)
http://dl.caffe.berkeleyvision.org/bvlc_googlenet.caffemodel
CNN_params.mean_file: absolute path to the '.mat' mean image for substraction
Semantic_params.api_key: API key of your IMAGGA account (for semantic features extraction)
Semantic_params.api_secret: API password of your IMAGGA account (for semantic features extraction)
demo.m
folder: absolute path to the lifelogging images we want to segment
format: format of the images in 'folder'
Just run to test on the demo dataset with the default parameters, or change
the parameters (in loadParamatersDemo.m) consequently.
The output of the demo function will be a .txt file in 'Results' folder, where each line describes the list
of images segmented in a different event. Additionally, you can generate three kinds of mosaics as an output
(see plot_params in loadParametersDemo.m).