基于改进CNN-LSTM的挖掘机作业对象识别

胡从裕,殷晨波,马伟,杨超,颜士宽

Object recognition of excavator operation based on improved CNN-LSTM

Congyu HU,Chenbo YIN,Wei MA,Chao YANG,Shikuan YAN

表 8 不同模型在各特征集上的性能指标

Tab.8 Performance indicator of different models on each feature set

模型 特征集 A P R F1 HBA-CNN S1 0.9308 0.9398 0.9383 0.9384 S2 0.9287 0.9223 0.9225 0.9224 S3 0.9325 0.9227 0.9323 0.9373 S4 0.8798 0.8791 0.8792 0.8896 HBA-GRU S1 0.8763 0.8877 0.8919 0.8762 S2 0.8621 0.8616 0.8620 0.8613 S3 0.8719 0.8716 0.8717 0.8712 S4 0.7443 0.7453 0.7442 0.7434 HBA-LSTM S1 0.9502 0.9560 0.9571 0.9566 S2 0.9330 0.9332 0.9347 0.9339 S3 0.9474 0.9407 0.9484 0.9445 S4 0.9289 0.9304 0.9352 0.9336 HBA-CNN- LSTM-Attention S1 0.9821 0.9811 0.9813 0.9812 S2 0.9784 0.9786 0.9773 0.9775 S3 0.9796 0.9781 0.9762 0.9771 S4 0.9243 0.9151 0.9202 0.9176