fruit_detection - Object detection

Dmitry / fruit_detection Public

Target: Himax WiseEye2 (Cortex-M55 400 MHz) + Ethos-U55-64 Clone this project

YOLOv5 is no longer available as a built-in block. Due to a policy change on hosting third party models, you must now upload this block yourself. Existing models still work (incl. deployment), but retraining requires following edgeimpulse/yolov5.

Object detection settings

Training settings

Training processor

Please provide a valid training processor option

Number of training cycles

Number of training cycles should be a number

Model size

Batch size

Batch size should be a number

Advanced training settings

Validation set size

Please provide a valid number for the train/validate split (between 0 and 1)

Split train/validation set on metadata key

Profile int8 model

Neural network architecture

Save

Input layer (49,152 features)

Community blocks / YOLOv5

Output layer (3 classes)

Model

Model version:

Last training performance (validation set)

mAP@50

0.88

Metrics (validation set)

Metric	Value
Precision score (legacy)	80.7%
mAP	0.48
mAP@[IoU=50]	0.88
mAP@[IoU=75]	0.54
mAP@[area=small]	0.13
mAP@[area=medium]	0.45
mAP@[area=large]	0.83
Recall@[max_detections=1]	0.39
Recall@[max_detections=10]	0.56
Recall@[max_detections=100]	0.56
Recall@[area=small]	0.16
Recall@[area=medium]	0.53
Recall@[area=large]	0.83

On-device performance

Last training performance (validation set)

mAP@50

0.89

Metrics (validation set)

Metric	Value
Precision score (legacy)	82.2%
mAP	0.52
mAP@[IoU=50]	0.89
mAP@[IoU=75]	0.62
mAP@[area=small]	0.23
mAP@[area=medium]	0.50
mAP@[area=large]	0.78
Recall@[max_detections=1]	0.39
Recall@[max_detections=10]	0.59
Recall@[max_detections=100]	0.61
Recall@[area=small]	0.34
Recall@[area=medium]	0.59
Recall@[area=large]	0.79

On-device performance

These users will be notified over email when this job finishes.

	Name
	D Dmitry

Object detection settings

Training settings

Augmentation settings

Advanced training settings

Neural network architecture

Model

Last training performance (validation set)

mAP@50

Confusion matrix (validation set)

Metrics (validation set)

Data explorer (full training set)

On-device performance

Last training performance (validation set)

mAP@50

Confusion matrix (validation set)

Metrics (validation set)

Data explorer (full training set)

On-device performance

Object detection settings

Training settings

Augmentation settings

Advanced training settings

Neural network architecture

Model

Last training performance (validation set)

mAP@50

Confusion matrix (validation set)

Metrics (validation set)

Data explorer (full training set)

Settings

On-device performance

Last training performance (validation set)

mAP@50

Confusion matrix (validation set)

Metrics (validation set)

Data explorer (full training set)

Settings

On-device performance