Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Our previous model was trained on images of size 1024. With the number of predictions that need to be made increasing on stations equipped with ptz cameras (smaller angle of view, therefore more images) and the reduction to 3 cpu, the pi doesn't have (almost) enough time to analyze everything. I'm also afraid of overuse. I therefore propose to switch to our best model in 640, reference 019aeef76fde49588d24fb167ba881a9.
We go from
On test test :
F1 : 0.916750 P : 0.934415 R : 0.899740 (threshold 0.1)
to
F1 : 0.890486 P : 0.933955 R : 0.850883 (threshold 0.1)
On test test :
F1 : 0.824006 P : 0.803797 R : 0.845258 (threshold 0.15)
to
F1 : 0.801754 P : 0.849442 R : 0.759136 (threshold 0.15)
I will re more training soon with 640 models only