step-counter

Setting up TFlite for Arduino

Follow the steps from this repository Clone examples repository to arduino library folder to make it available in Arduino IDE and include TensorFlowLite.h in your sketch.

Metrics

Path	num_samples	test.f1-score_macro	test.precision_macro	test.predicted_step_count	test.recall_macro	test.roc_auc	test.support_macro	test.target_step_count	train.accuracy	train.f1_macro	train.precision_macro	train.recall_macro	validation.accuracy	validation.f1_macro	validation.precision_macro	validation.recall_macro	validation.threshold_mean	validation.threshold_std
reports/metrics/train.json	7919	-	-	-	-	-	-	-	0.87	0.79	0.76	0.85	0.9	0.8	0.8	0.8	0.7	0.05
reports/metrics/test.json	-	0.82	0.83	53	0.81	0.92	1585.0	45	-	-	-	-	-	-	-	-	-	-
reports/metrics_tflite/test.json	-	0.82	0.83	53	0.81	0.92	1585.0	45	-	-	-	-	-	-	-	-	-	-

Steps Graph

flowchart TD
	node1["convert_tflite_to_c"]
	node2["data/steps/test.dvc"]
	node3["data/steps/train.dvc"]
	node4["evaluate"]
	node5["evaluate_tflite"]
	node6["export_tflite"]
	node7["predict_test"]
	node8["predict_tflite"]
	node9["train"]
	node10["update-metrics"]
	node2-->node4
	node2-->node5
	node2-->node7
	node2-->node8
	node3-->node6
	node3-->node9
	node4-->node10
	node5-->node10
	node6-->node1
	node6-->node8
	node7-->node4
	node7-->node10
	node8-->node5
	node8-->node10
	node9-->node6
	node9-->node7
	node9-->node10

graph_end