Lightweight Strawberry Recognition with Hybrid Deep Deformable
Convolution and Double Collaborative Attention Mechanisms
- Fengqian Pang,
- Xi Chen
Abstract
The existing ripeness detection algorithm for strawberries suffers from
low detection accuracy and high detection error rate. Considering these
problems, we propose an improvement method based on YOLOv5, which
firstly reconfigures the feature extraction network by replacing
ordinary convolution with hybrid depth deformable convolution. In the
second step, a double cooperative attention mechanism is constructed to
improve the representation of strawberry features in complex
environments. Finally, cross-scale feature fusion is proposed to fully
integrate the multiscale target features. The method was tested on the
strawberry ripeness dataset, the mAP reached 95.6 percentage points, the
FPS reached 76, and the model size was 7.44M. The mAP and FPS are 8.4
and 1.3 percentage points higher respectively than the baseline network.
The model size is reduced by 6.28M. This method is superior to many
state-of-the-art algorithms in terms of detection speed and accuracy.
The system can accurately identify the ripeness of strawberries in
complex environments, which could provide technical support for
automated picking robots.14 Sep 2023Submitted to The Journal of Engineering 15 Sep 2023Submission Checks Completed
15 Sep 2023Assigned to Editor
27 Sep 2023Reviewer(s) Assigned
25 Oct 2023Review(s) Completed, Editorial Evaluation Pending
29 Oct 2023Editorial Decision: Revise Minor
09 Nov 20231st Revision Received
11 Nov 2023Submission Checks Completed
11 Nov 2023Assigned to Editor
11 Nov 2023Reviewer(s) Assigned
22 Nov 2023Review(s) Completed, Editorial Evaluation Pending
22 Nov 2023Editorial Decision: Accept