DIODE: Dilatable Incremental Object Detection
To accommodate rapid changes in the real world, the cognition system of humans is capable of continually learning concepts. On the contrary, conventional deep learning models lack this capability of preserving previously learned knowledge. When a neural network is fine-tuned to learn new tasks, its...
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Artículo |
| Publicado: |
2021
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2108.05627 http://arxiv.org/abs/2108.05627 |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|