Detection of Package Edges in Distance Maps

Vasileva, Elena; Avramovski, Nenad; Ivanovski, Zoran

doi:10.23919/eusipco47968.2020.9287558

Cited by 2 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this paper, we propose an encoderdecoder CNN with a limited number of parameters, but enough capacity to retain the crucial dataset features for correct segmentation of box edges. This CNN, presented in our previous work [62], is briefly described in Section IV.C.…”

Section: Edge Detectionmentioning

confidence: 99%

“…Each connected component in the binary mask of box sides represents one side of a package. Another end-to-end trainable CNN [62] with a similar structure performs segmentation of box edges as connected components. A custom edge thinning algorithm based on non-maximum suppression generates binary masks of edges with 1 pixel thickness.…”

Section: Package Detectionmentioning

confidence: 99%

“…To obtain precise localization of the package borders and more accurate segmentation in the areas around rounded inner edges, we perform edge detection in the distance maps using a CNN-based edge detection algorithm that consists of a CNN optimized for training with a limited number of samples, followed by a custom edge thinning algorithm. The edge detection algorithm is described in our previous work [62]. Since the inner and outer edges are marked by significantly different characteristics (as discussed in detail in Section IV.A), the CNN performs classification of inner and outer package edges into two separate classes, thus reducing the in-class variance, and simplifying the problem of detecting rounded inner edges.…”

Section: Detection Of Package Edgesmentioning

confidence: 99%

“…The algorithm correctly segments box edges represented by a sufficient number of distance points in the presence of heavy distortion in the distance maps. More details on the edge detection algorithm and performance metrics can be seen in [62].…”

Section: B Segmentation Of Package Edgesmentioning

confidence: 99%

See 3 more Smart Citations

A Hybrid CNN-DSP Algorithm for Package Detection in Distance Maps

Vasileva,

Ivanovski

2023

IEEE Access

Self Cite

View full text Add to dashboard Cite

This paper presents a hybrid algorithm for real-time instance segmentation of packages from scenes represented by 2D distance maps (range images). The paper introduces a novel approach combining deep learning-based methods and digital signal processing methods to enable accurate package recognition, using a small training dataset with high variability and distance measurement errors characteristic of Timeof-Flight-based scanning. Two convolutional neural networks with architecture optimized for training with a limited number of samples perform an initial segmentation of package components (sides and edges). An algorithm based on digital signal processing methods performs refinement of intermediate results, and combines package components into packages. Training and evaluation of the algorithm were performed on a custom dataset containing scenes of packages, shipping bags, and packaging of irregular shapes with various sizes, orientations, and degrees of occlusion, organized either in ordered stacks or arbitrary order. The convolutional neural networks provide a reliable distinction between components of packages and components of other types of packaging and surroundings. Package sides containing a sufficient number of distance points are correctly combined into packages. Thus, the proposed algorithm represents a solid basis for fully automated loading/unloading of packages with arbitrary sizes and materials from transport trailers and storage spaces. The dataset and annotations for box side surfaces are available at: https://dipteam.feit.ukim.edu.mk/results-package-detection.html

show abstract

Section: Edge Detectionmentioning

confidence: 99%

Section: Package Detectionmentioning

confidence: 99%

Section: Detection Of Package Edgesmentioning

confidence: 99%

Section: B Segmentation Of Package Edgesmentioning

confidence: 99%

See 2 more Smart Citations