Verification and calculation of map, precision and recall values in roboflow

  1. Project type (Object detection)

I have finished training the data and export the model to roboflow however how to verify the result of map, precision and recall values in google colab?

Hi @Racheal_c

Could you clarify what you mean by verifying the results?

In Roboflow, you can check your mAP, Precision and Recall statistics by clicking on the “Details” button on each trained version page or scrolling down:

In Colab once you’re done training, you can view your trained model stats on the notebook. Where it shows up depends on which notebook you’re running, but usually they’ll show up like this:

Thanks for your reply. Verifying the results means the image you attached here from google colab .

But for my notebook I didn’t get this, I’m using these codes from this notebook Google Colab

I only getting this (attached below) and does not have the summary of mAP, recall and precision

Do you know what I can do to add to have the results like the one you attached from google colab?

Hi @Racheal_c

On Google Colab, you can see a lot of the result stats you are looking for here under content/runs/detect/train:

If mAP is all you’re looking for, you can find it in the table for the model summary that is sent after it’s finished training here:
It has mAP for each class as well as for all of them combined.

If you are looking for precision and recall statistics, the results.png displayed in the line in the notebook will have recall and precision in the charts.


If you need exact numerical data, it can be found in a CSV file at content/runs/detect/train/results.csv.

Alternatively, an easier solution is deploying your model on Roboflow:

It will show you mAP, precision and recall information and give you a API you can directly infer with.

Hope this helps