Boundary-aware small object detection with attention and interaction

Feng, Qihan; Shao, Zhiwen; Wang, Zhixiao

doi:10.1007/s00371-023-03144-x

Cited by 2 publications

(3 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In step 1, for each pixel in the image, directional matrices H 45° and H 135° were added for expanding and refining gradient direction based on directional matrices H x and H y of the classical Canny algorithm. The specific calculation process is shown in Equations ( 10) and (11). M i,j = H x i,j 2 +H y i,j 2 +H 45° i,j 2 +H 135° i,j 2 (10)…”

Section: Edge Extraction and Peak Detection Counting Methods 241 Edge...mentioning

confidence: 99%

“…The images captured by drones have the problem of many targets to be detected and small target sizes. The main solution is to increase the number of detection pyramid layers; however, this will weaken the representation of details in low-level features and increase the model size [11]. Deep learning can learn the growth patterns of crops, and efficiently and accurately identify the types and positions of crops, and is widely used in the field of digital image recognition [12].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Assisting the Planning of Harvesting Plans for Large Strawberry Fields through Image-Processing Method Based on Deep Learning

Wang,

Han,

et al. 2024

Agriculture

View full text Add to dashboard Cite

Reasonably formulating the strawberry harvesting sequence can improve the quality of harvested strawberries and reduce strawberry decay. Growth information based on drone image processing can assist the strawberry harvesting, however, it is still a challenge to develop a reliable method for object identification in drone images. This study proposed a deep learning method, including an improved YOLOv8 model and a new image-processing framework, which could accurately and comprehensively identify mature strawberries, immature strawberries, and strawberry flowers in drone images. The improved YOLOv8 model used the shuffle attention block and the VoV–GSCSP block to enhance identification accuracy and detection speed. The environmental stability-based region segmentation was used to extract the strawberry plant area (including fruits, stems, and leaves). Edge extraction and peak detection were used to estimate the number of strawberry plants. Based on the number of strawberry plants and the distribution of mature strawberries, we draw a growth chart of strawberries (reflecting the urgency of picking in different regions). The experiment showed that the improved YOLOv8 model demonstrated an average accuracy of 82.50% in identifying immature strawberries, 87.40% for mature ones, and 82.90% for strawberry flowers in drone images. The model exhibited an average detection speed of 6.2 ms and a model size of 20.1 MB. The proposed new image-processing technique estimated the number of strawberry plants in a total of 100 images. The bias of the error for images captured at a height of 2 m is 1.1200, and the rmse is 1.3565; The bias of the error for the images captured at a height of 3 m is 2.8400, and the rmse is 3.0199. The assessment of picking priorities for various regions of the strawberry field in this study yielded an average accuracy of 80.53%, based on those provided by 10 experts. By capturing images throughout the entire growth cycle, we can calculate the harvest index for different regions. This means farmers can not only obtain overall ripeness information of strawberries in different regions but also adjust agricultural strategies based on the harvest index to improve both the quantity and quality of fruit set on strawberry plants, as well as plan the harvesting sequence for high-quality strawberry yields.

show abstract

Section: Edge Extraction and Peak Detection Counting Methods 241 Edge...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Assisting the Planning of Harvesting Plans for Large Strawberry Fields through Image-Processing Method Based on Deep Learning

Wang,

Han,

et al. 2024

Agriculture

View full text Add to dashboard Cite

show abstract

“…In order to enhance the precision of object detection, we strive to improve the widely used YOLOv7 algorithm, which is depicted in Figure 2 as its network structure. There are many improved networks based on the YOLO network, and some are improved by using a single module [37][38][39], some are embedded in the network based on the large model [40][41][42], and some add new structures to the original network [43]. At the same time, there are also many improvements in small object detection based on the YOLO series in different application scenarios [44][45][46][47].…”

Section: Improvement Of Network Structurementioning

confidence: 99%

A Small Object Detection Method for Drone-Captured Images Based on Improved YOLOv7

Zhao,

Shao,

Liu

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Due to the broad usage and widespread popularity of drones, the demand for a more accurate object detection algorithm for images captured by drone platforms has become increasingly urgent. This article addresses this issue by first analyzing the unique characteristics of datasets related to drones. We then select the widely used YOLOv7 algorithm as the foundation and conduct a comprehensive analysis of its limitations, proposing a targeted solution. In order to enhance the network’s ability to extract features from small objects, we introduce non-strided convolution modules and integrate modules that utilize attention mechanism principles into the baseline network. Additionally, we improve the semantic information expression for small targets by optimizing the feature fusion process in the network. During training, we adopt the latest Lion optimizer and MPDIoU loss to further boost the overall performance of the network. The improved network achieves impressive results, with mAP50 scores of 56.8% and 94.6% on the VisDrone2019 and NWPU VHR-10 datasets, respectively, particularly in detecting small objects.

show abstract

Boundary-aware small object detection with attention and interaction

Cited by 2 publications

References 40 publications

Assisting the Planning of Harvesting Plans for Large Strawberry Fields through Image-Processing Method Based on Deep Learning

Assisting the Planning of Harvesting Plans for Large Strawberry Fields through Image-Processing Method Based on Deep Learning

A Small Object Detection Method for Drone-Captured Images Based on Improved YOLOv7

Contact Info

Product

Resources

About