Research on Coal Gangue Detection and Recognition Based on Lightweight Network MS-YOLOV3

• Autorzy: Li Deyong , Wang Guofa , Wang Shuang , Wang Wenshan , Du Ming

Identyfikatory

DOI

10.24425/gsm.2022.143628

Warianty tytułu

PL

Badania nad wykrywaniem i rozpoznawaniem skały płonnej w oparciu o lekką sieć MS -YOLOV3

• Języki publikacji: EN

Abstrakty

EN

The rapid and accurate detection and identification of coal gangue is one of the premises and key technologies of the intelligent separation of coal gangue, which is of considerable importance for the separation of coal gangue. Focusing on the problems in the current deep learning algorithms for the detection and recognition of coal gangue, such as large model memory and slow detection speed, a rapid detection method for lightweight coal gangue is proposed. YOLOv3 is taken as the basic structure and improved. The MobileNetv2 lightweight feature extraction network is selected to replace Darknet53 as the main network of the detection algorithm to improve the detection speed. Spatial pyramid pooling (SPP) is added after the backbone network to convert different feature maps into fixed feature maps in order to improve the positioning accuracy and detection capability of the algorithm, thereby obtaining the lightweight network MS-YOLOV3. The experimental equipment was set up and multi-condition coal and gangue datasets were constructed. The model was trained and the identification and positioning results of the model were tested under different sizes, illumination intensities and various working conditions, and compared with other algorithms. Experimental results show that the proposed algorithm can detect the coal gangue quickly and accurately, with an mAP of 99.08%, a speed of 139 fps and a memory occupation of only 9.2 M. In addition, the algorithm can effectively detect mutually stacking coal and gangue of different quantities and sizes under different lights with high confidence and with a certain degree of environmental robustness and practicability. Compared with the YOLOv3, the performance of the proposed algorithm is significantly improved. Under the premise that the accuracy is unchanged, the FPS increases by 127.9% and the memory decreases by 96.2%. Therefore, the MS-YOLOv3 algorithm has the advantages of small memory, high accuracy and fast speed, which can provide online technical support for the detection and identification of coal and gangue.

PL

Szybkie i dokładne wykrywanie oraz identyfikacja skały płonnej jest jedną z przesłanek i kluczowych technologii inteligentnej separacji skały płonnej. Koncentrując się na problemach związanych z obecnymi algorytmami wykrywania i rozpoznawania skały płonnej z głębokim uczeniem, takimi jak duża pamięć modelu i niska prędkość wykrywania, zaproponowano metodę szybkiego wykrywania lekkiej skały płonnej. YOLOv3 jest traktowany jako struktura podstawowa i ulepszony. Lekka sieć ekstrakcji funkcji Mobilenetv2 została wybrana w celu zastąpienia Darknet53 jako głównej sieci algorytmu wykrywania w celu poprawy szybkości wykrywania. Spatial Pyramid Pooling (SPP) jest dodawany po sieci szkieletowej w celu konwersji różnych map obiektów na mapy stałych funkcji, aby poprawić dokładność pozycjonowania i zdolność wykrywania algorytmu, uzyskując w ten sposób lekką sieć MS-YOLOV3. Ustawiono sprzęt eksperymentalny i skonstruowano wielowarunkowe zbiory danych dotyczące węgla i skały płonnej. Model został przeszkolony, a wyniki identyfikacji i pozycjonowania modelu zostały przetestowane przy różnych rozmiarach, natężeniu oświetlenia i różnych warunkach pracy oraz porównane z innymi algorytmami. Wyniki eksperymentu pokazują, że zaproponowany algorytm jest w stanie szybko i dokładnie wykryć skałę węglową, z mAP na poziomie 99,08%, szybkością 139 fps i zajęciem pamięci zaledwie 9,2 MB. Ponadto może skutecznie wykrywać różne światła, różne rozmiary, wzajemne układanie w stosy oraz wielokrotną ilość węgla i skały płonnej, z dużą pewnością i pewną odpornością środowiskową i wykonalnością. W porównaniu z YOLOv3 wydajność proponowanego algorytmu jest znacznie lepsza. Przy założeniu, że dokładność pozostaje w zasadzie niezmieniona, FPS wzrasta o 127,9%, a pamięć spada o 96,2%. Dlatego algorytm MS-YOLOv3 ma zalety małej pamięci, wysokiej dokładności i dużej szybkości, co może zapewnić wsparcie techniczne dla wykrywania i identyfikacji węgla i skały płonnej online.

Słowa kluczowe

EN

coal gangue; MobileNetV2; spatial pyramid pooling; SPP; YOLOv3

PL

skała płonna; MobileNetV2; SPP; YOLOv3

• Wydawca: Instytut Gospodarki Surowcami Mineralnymi i Energią PAN ; Komitet Zrównoważonej Gospodarki Surowcami Mineralnymi PAN
• Czasopismo: Gospodarka Surowcami Mineralnymi
• Rocznik: 2022
• Tom: T. 38, z. 4
• Strony: 133--152
• Opis fizyczny: Bibliogr. 34 poz., rys., tab., wykr.

Twórcy

autor

Li Deyong

• State Key Laboratory of Mining Response and Disaster Prevention and Control in Deep Coal Mines, Anhui University of Science and Technology, Huainan 232001, China

autor

Wang Guofa

• Collaborative Innovation Center for Mine Intelligent Technology and Equipment, Anhui University of Science and Technology, Huainan 232001, China

• https://orcid.org/0000-0002-1772-6049

autor

Wang Shuang

• China Coal Technology Engineering Group Coal Mining Research Institute, Beijing 100013, China

autor

Wang Wenshan

• China Coal Technology Engineering Group China Coal Research Institute, Beijing 100013, China

autor

Du Ming

• China Coal Technology Engineering Group China Coal Research Institute, Beijing 100013, China

Bibliografia

• [1] Alfarzaeai et al. 2020 – Alfarzaeai, M., Niu, Q., Zhao, J., Eshaq, R. and Hu, E. 2020. Coal/Gangue Recognition Using Convolutional Neural Networks and Thermal Images. IEEE Access 8, pp. 76780–76789.
• [2] Bochkovskiy et al. 2020 – Bochkovskiy, A., Wang, C. and Liao, H. 2020. YOLOv4: Optimal Speed and Accuracy of Object Detection. arXiv preprint, arXiv:2004.10934.
• [3] Cao et al. 2022 – Cao, X., Liu, S., Wang, P., Xu, G. and Wu, D. 2022. Research on coal gangue identification and positioning system based on coal-gangue sorting robot. Coal Science and Technology 50(1), pp. 237–246 (in Chinese).
• [4] Cong et al. 2021 – Cong, L. and Xingru, W. 2021. Research and Practice of China’s Intelligent Coal Mines. [In:] In 2021 International Conference on Computer EngIneerIng and Artificial Intelligence (ICCEAI), pp. 391–399.
• [5] Guo et al. 2019 – Guo, Y., Yu Z. and Lu, Y. 2019. Research on photoelectric intelligent separation technology of coal and gangue based on NP-FSVM with the PSO algorithm. Coal Science and Technology 47(4), pp. 13–19 (in Chinese).
• [6] Guo et al. 2022 – Guo, Y., Wang, X., He, L. and Liu P. 2022. Research on Coal and Gangue Recognition Method Based on TW-RN Optimized CNN. Coal Science and Technology 50(1), pp. 228–236 (in Chinese).
• [7] Gupta, N. 2016. Evaluation of pneumatic inclined deck separator for high-ash Indian coals. International Journal of Coal Science – Technology 3(2), pp. 198–205, DOI: 10.1007/s40789-016-0125-2.
• [8] Haase et al. 2020 – Haase, D. and Amthor, M. 2020. Rethinking Depthwise Separable Convolutions: How Intra-Kernel Correlations Lead to Improved MobileNets.2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
• [9] He et al. 2015 – He, K., Zhang, X., Ren, S. and Sun, J. 2015. Spatial pyramid pooling in deep convolutional networks for visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence 37(9), pp.1904–1916.
• [10] He et al. 2022 – He, L., Wang, S., Guo, Y., Cheng, G., Hu, K., Zhao, Y. and Wang, X. 2022. Multi-scale coal and gangue dual-energy X-ray image concave point detection and segmentation algorithm. Measurement 196, DOI: 10.1016/j.measurement.2022.111041.
• [11] Hou, W. 2017. Identification of coal and gangue by feed-forward neural network based on data analysis. International Journal of Coal Preparation and Utilization 39(1), pp. 33–43, DOI: 10.1080/19392699.2017.1290609.
• [12] Howard et al. 2017 – Howard, A., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M. and Adam, H. 2017. MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications, DOI: 10.48550/arXiv.1704.04861.
• [13] Klojzy-Karczmarczyk et al.2016 – Klojzy-Karczmarczyk, B., Mazurek, J., and Paw, K. 2016. Possibilities of Utilization of Aggregates and Extractive Waste from hard Coal Mining at Janina Mine in the Process of Reclamation of Open-pit Mines. Gospodarka Surowcami Mineralnymi – Mineral Resources Management 32(3), pp. 111–133, DOI: 10.1515/gospo-2016-0030.
• [14] Li et al. 2020 – Li, M., Duan, Y., Cao, X., Liu, C., Sun, K. and Liu, H. 2019. Image recognition method and system of coal gangue sorting robot. Journal of China Coal Society 46(9), pp. 1305–1315, DOI: 10.1080/19392699. 2020.1760855 (in Chinese).
• [15] Li, Y. 2020. Research on coal gangue detection based on deep learning. Xi’an University of Science and Technology (in Chinese).
• [16] Li et al. 2021 – Li, D., Wang, G., Zhang, Y. and Wang, S.2021. Coal gangue detection and recognition algorithm based on deformable convolution YOLOv3. IET Image Process 16(1), pp. 134–144, DOI: 10.1049/ipr2.12339.
• [17] Li et al. 2022 – Li, D., Wang, G., Guo, Y., Zhang, Y. and Wang, S.2022. An identification and positioning method for coal gangue based on lightweight mixed domain attention. International Journal of Coal Preparation and Utilization, DOI: 10.1080/19392699.2022.2119561.
• [18] Lin et al. 2017 – Lin, T., Dollar, P., Girshick, R., He, K., Hariharan, B. and Belongie, S. 2017. Feature Pyramid Networks for Object Detection. 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). IEEE Computer Society.
• [19] Liu et al. 2019 – Liu, C., Li, M., Zhang, Y., Han, S., and Zhu, Y. 2019. An Enhanced Rock Mineral Recognition Method Integrating a Deep Learning Model and Clustering Algorithm. Minerals 9(9), DOI: 10.3390/min9090516.
• [20] Lv et al. 2022 – Lv, Z., Wang, W., Zhang, K., Li, W., Feng, J. and Xu, Z.2022. Asynchronous detection-segmentation method for oversized gangue on a coal preparation plant based on multi-task learning. Minerals Engineering 187, DOI: 10.1016/j.mineng.2022.107806.
• [21] Mccoy, J. and Auret, L. 2019. Machine learning applications in minerals processing: Areview. Minerals Engineering 132, pp. 95–109, DOI: 10.1016/j.mineng.2018.12.004.
• [22] Pan et al. 2022 – Pan, H., Shi, Y., Lei, X., Wang, Z. and Xin, F., 2022. Fast identification model for coal and gangue based on the improved tiny YOLOv3. J. Real-Time Image Process 19(3), pp. 687–701, DOI: 10.1007/s11554-022-01215-1.
• [23] Pu et al. 2019 – Pu, Y., Apel, D., Szmigiel, A. and Chen, J. 2019. Image Recognition of Coal and Coal Gangue Using a Convolutional Neural Network and Transfer Learning. Energies 12(9), DOI: 10.3390/en12091735.
• [24] Redmon, J. and Farhadi, A. 2018. YOLOv3: An Incremental Improvement. arXiv e-prints. DOI: 10.48550/arXiv.1804.02767.
• [25] Sahu et al. 2017 – Sahu, L. and Dey, S. 2017. Enrichment of carbon recovery of high ash coal fines using air fluidized vibratory deck separator. International Journal of Coal Science – Technology 4(3), pp. 262–273, DOI: 10.1007/s40789-017-0172-3.
• [26] Sandler et al.2018 – Sandler, M., Howard, A., Zhu, M., Zhmoginov, A. and Chen, L. 2018. MobileNetV2: Inverted Residuals and Linear Bottlenecks. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).
• [27] Wang et al. 2018 – Wang, J., Li, L. and Yang, S. 2018. Experimental study on gray and texture features extraction of coal and gangue image under different illuminance. Journal of China Coal Society 43(11), pp. 3051–3061, DOI: 10.13225/j.cnki.jccs.2018.0866 (in Chinese).
• [28] Wang et al. 2019a – Wang, G., Liu, F., Meng, X., Fan, J., Wu, Q., Ren, H., Pang, Y., Xu, Y., Zhao, G., Zhang, D., Cao, X., Du, Y., Zhang, J., Chen, H., Ma, Y. and Zhang, K. 2019a. Research and practice on intelligent coal mine construction (primary stage). Coal Science and Technology 47(8), pp. 1–36 (in Chinese).
• [29] Wang et al. 2019b – Wang, G., Liu, F., Pang, Y., Ren, H. and Ma, Y. 2019b.Coal mine intellectualization: The core technology of high quality development. Journal of China Coal Society 44(2), pp. 349–357 (in Chinese).
• [30] Wang et al. 2021 – Wang, X., Wang, S., Guo, Y., Hu, K. and Wang, W. 2021. Dielectric and geometric feature extraction and recognition method of coal and gangue based on VMD-SVM. Powder Technology 392, pp. 241–250, DOI: 10.1016/j.powtec.2021.06.057.
• [31] Zhang, N. 2015. Detection and radiation law of natural gamma ray from coal and roof-rock in the fully mechanized top coal caving mining. Xuzhou: China University of Mining – Technology (in Chinese).
• [32] Zhang et al. 2020 – Zhang, Z., Liu, Y., Hu, Q., Zhang, Z., Wang, L., Liu, X. and Xia, X. 2020. Multi-information online detection of coal quality based on machine vision. Powder Technology 374, pp. 250–262, DOI: 10.1016/j.powtec.2020.07.040.
• [33] Zhao et al. 2014 – Zhao, Y., Yang, X., Luo, Z., Duan, Z. and Song, S. 2014. Progress in developments of dry coal beneficiation. International Journal of Coal Science – Technology 1(1), pp. 103–112, DOI: 10.1007/s40789-014-0014-5.
• [34] Zhao et al. 2022 – Zhao, Y., Wang, S., Guo, Y., Cheng, G., He, L., and Wang, W. 2022. The identification of coal and gangue and the prediction of the degree of coal metamorphism based on the EDXRD principle and the PSOSVM model. Gospodarka Surowcami Mineralnymi – Mineral Resources Management 38(2), pp. 113–129, DOI: 10.24425/gsm.2022.141663.

Uwagi

PL

Opracowanie rekordu ze środków MEiN, umowa nr SONP/SP/546092/2022 w ramach programu "Społeczna odpowiedzialność nauki" - moduł: Popularyzacja nauki i promocja sportu (2022-2023)