Instructions
You can use this demo to predict situations for images of your choice.
Start by clicking on the example images.
The demo provides the nearest neighbors in the imSitu training set and a list of predicted situations and associated probability.
The demo has a 2 second time limit, so please do not try to very large images because they will take the server too long to download.
The demo is an implemenation of the Compositional Conditional Random Field, pre-trained using semantic data augmention on 5 million web images, found here.
Click on images to try these examples
Recognize Situations
Paste a url
Upload a file
Results
Query
Similar in imSitu train set
Predicted situations
| fixing | ||||
| agent | object | objectpart | tool | place |
| man | roof | shingle | hand | outside |
0.30590
| repairing | ||||
| agent | item | problem | tool | place |
| man | roof | shingle | hand | outside |
0.23353
| constructing | ||||
| agent | createditem | tool | components | place |
| man | roof | ∅ | beam | outdoors |
0.13213
| mending | |||
| agent | item | tool | place |
| man | roof | hand | roof |
0.09530
| painting | |||
| agent | item | tool | place |
| man | roof | brush | outdoors |
0.06129
| stripping | ||||
| agent | removeditem | source | tool | place |
| man | shingle | roof | hand | rooftop |
0.05668
| installing | ||||
| agent | component | destination | tool | place |
| man | solar array | roof | hand | outdoors |
0.04116
| building | ||||
| agent | goalitem | components | tool | place |
| man | roof | wood | hand | outdoors |
0.02805
| nailing | ||||
| agent | item1 | item2 | tool | place |
| man | board | roof | machine | outside |
0.01684
| tearing | |||
| agent | item | tool | place |
| man | roof | sword | outdoors |
0.00580