Difference between revisions of "DSC180F20W10"
From Immersive Visualization Lab Wiki
Line 6: | Line 6: | ||
* Summary of replication result | * Summary of replication result | ||
+ | ** Include at least two images: one in which the classification works well and the GradCAM images confirm it. And one where the classification is wrong and GradCAM helps to know why. | ||
* Shortcomings | * Shortcomings | ||
* Possible improvements | * Possible improvements |
Revision as of 22:14, 13 December 2020
Final Replication Checkpoint
The goal of the final project is to apply the Grad-CAM algorithm to the COCO dataset.
Report
- Summary of replication result
- Include at least two images: one in which the classification works well and the GradCAM images confirm it. And one where the classification is wrong and GradCAM helps to know why.
- Shortcomings
- Possible improvements
- Conclusions
Code
- Develop functional Grad-CAM code which uses the COCO data set
- Develop code that generates an analysis and figures for results
- Use this code as a basis