|
|
--- |
|
|
tags: |
|
|
- fastai |
|
|
--- |
|
|
## Model description |
|
|
This repo contains the trained model for Multi-object classification |
|
|
|
|
|
Full credits go to [Nhu Hoang](https://www.linkedin.com/in/nhu-hoang/) |
|
|
|
|
|
Motivation: Classifying multiple objects is a challenging task without using an object detection algorithm. This model was trained on resnet34 backbone and achieved a good accuracy. |
|
|
|
|
|
## Training and evaluation data |
|
|
### Training hyperparameters |
|
|
|
|
|
The following hyperparameters were used during training: |
|
|
|
|
|
| Hyperparameters | Value | |
|
|
| :-- | :-- | |
|
|
| name | Adam | |
|
|
| learning_rate | 3e-3 | |
|
|
| training_precision | float16 | |
|
|
|
