|
85 | 85 | "id": "PzoNZRp4sVxK" |
86 | 86 | }, |
87 | 87 | "source": [ |
88 | | - "If you want to run the notebook with the salad dataset, you can run the whole thing now by clicking **Runtime \u003e Run all** in the Colab toolbar. But if you want to use your own dataset, then continue down to [Load the training data](#scrollTo=H0XM-oIfhgQ7) and follow the instructions there." |
| 88 | + "If you want to run the notebook with the salad dataset, you can run the whole thing now by clicking **Runtime \u003e Run all** in the Colab toolbar. But if you want to use your own dataset, then continue down to [Load the training data](#scrollTo=H0XM-oIfhgQ7) and follow the instructions there.\n", |
| 89 | + "\n", |
| 90 | + "**Note:** If using a custom dataset, beware that if your dataset includes more than 20 classes, you'll probably have slower inference speeds compared to if you have fewer classes. This is due to an aspect of the EfficientDet architecture in which a certain layer cannot compile for the Edge TPU when it carries more than 20 classes." |
89 | 91 | ] |
90 | 92 | }, |
91 | 93 | { |
|
0 commit comments