Merged
Conversation
Wrap the production pyro-predictor package (YOLO ONNX + sliding-window temporal smoothing) as a TemporalModel baseline. Uses production defaults from pyro-engine's Engine class (conf_thresh=0.35, nb_consecutive_frames=7). Dataset imported from pyro-dataset v2.2.0 via dvc import. DVC pipeline: prepare -> predict (train/val) -> evaluate.
…b_consecutive_frames=7)
….0 model - Add package module and script to bundle ONNX weights + config into a deployable .zip archive (mirrors FSM baseline pattern) - Refactor predict script to use PyroDetectorModel (TemporalModel subclass) instead of calling predictor_wrapper directly - Add from_package classmethod to PyroDetectorModel for loading archives - Switch YOLO model to pyronear/yolo11s_nimble-narwhal_v6.0.0 - Add 10 package tests (build, roundtrip, error cases)
…ections Split YOLO inference from temporal logic so the model runs once: - infer stage: runs YOLO Classifier, caches per-frame detections as JSON - predict stage: replays cached detections through Predictor temporal logic - sweep stage: grid-searches conf_thresh x nb_consecutive_frames (64 combos) Best val F1=0.913 at conf=0.2, nb_frames=4 (vs 0.825 with production defaults).
Sweep found best val F1=0.913 at conf_thresh=0.2, nb_consecutive_frames=4 (vs F1=0.825 with production defaults conf=0.35, nb_frames=7).
Move create_replay_predictor(), load_detections(), and replay_sequence() from predict.py and sweep.py into predictor_wrapper.py to eliminate duplication. Update DVC deps to include predictor_wrapper.py.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
pyro-predictorpackage (YOLO ONNX + sliding-window temporal smoothing) as aTemporalModelbaselineEngineclass (conf_thresh=0.35,nb_consecutive_frames=7)dvc import(sequential_train_val split)prepare->predict(train/val) ->evaluate(4 slices)Test plan
make lintpassesmake testpasses (41 tests)uv run dvc reprocompletes successfullyuv run dvc metrics show)