Deep inside convolutional networks: visualising image classification models and saliency maps
This paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises...
| Main Authors: | , , |
|---|---|
| Format: | Conference item |
| Language: | English |
| Published: |
International Conference on Learning Representations
2014
|
| _version_ | 1826315165760487424 |
|---|---|
| author | Simonyan, K Vedaldi, A Zisserman, A |
| author_facet | Simonyan, K Vedaldi, A Zisserman, A |
| author_sort | Simonyan, K |
| collection | OXFORD |
| description | This paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises the class score [5], thus visualising the notion of the class, captured by a ConvNet. The second technique computes a class saliency map, specific to a given image and class. We show that such maps can be employed for weakly supervised object segmentation using classification ConvNets. Finally, we establish the connection between the gradient-based ConvNet visualisation methods and deconvolutional networks [13]. |
| first_indexed | 2024-12-09T03:20:42Z |
| format | Conference item |
| id | oxford-uuid:c46c7936-e692-4f58-b809-1e9e471f220e |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:20:42Z |
| publishDate | 2014 |
| publisher | International Conference on Learning Representations |
| record_format | dspace |
| spelling | oxford-uuid:c46c7936-e692-4f58-b809-1e9e471f220e2024-11-05T13:00:38ZDeep inside convolutional networks: visualising image classification models and saliency mapsConference itemhttp://purl.org/coar/resource_type/c_5794uuid:c46c7936-e692-4f58-b809-1e9e471f220eEnglishSymplectic ElementsInternational Conference on Learning Representations2014Simonyan, KVedaldi, AZisserman, AThis paper addresses the visualisation of image classification models, learnt using deep Convolutional Networks (ConvNets). We consider two visualisation techniques, based on computing the gradient of the class score with respect to the input image. The first one generates an image, which maximises the class score [5], thus visualising the notion of the class, captured by a ConvNet. The second technique computes a class saliency map, specific to a given image and class. We show that such maps can be employed for weakly supervised object segmentation using classification ConvNets. Finally, we establish the connection between the gradient-based ConvNet visualisation methods and deconvolutional networks [13]. |
| spellingShingle | Simonyan, K Vedaldi, A Zisserman, A Deep inside convolutional networks: visualising image classification models and saliency maps |
| title | Deep inside convolutional networks: visualising image classification models and saliency maps |
| title_full | Deep inside convolutional networks: visualising image classification models and saliency maps |
| title_fullStr | Deep inside convolutional networks: visualising image classification models and saliency maps |
| title_full_unstemmed | Deep inside convolutional networks: visualising image classification models and saliency maps |
| title_short | Deep inside convolutional networks: visualising image classification models and saliency maps |
| title_sort | deep inside convolutional networks visualising image classification models and saliency maps |
| work_keys_str_mv | AT simonyank deepinsideconvolutionalnetworksvisualisingimageclassificationmodelsandsaliencymaps AT vedaldia deepinsideconvolutionalnetworksvisualisingimageclassificationmodelsandsaliencymaps AT zissermana deepinsideconvolutionalnetworksvisualisingimageclassificationmodelsandsaliencymaps |