6
• v1 score =96.36%
Comparing the learning accuracy in Figure 7 with the
testing accuracy in Figure 8, the proposed does not suf-
fer from overfitting or under-fitting which means that
the proposed model is robust and can be used in differ-
ent underground mines. The model succeeded to segment
color images that have been captured by a traditional cam-
era and classify the pixels that belong to rocks from pixels
that belong to straps. In addition, the model could locate
the holes in the straps which belong to rock.
Figure 9 shows the testing of the model for the scenario
where the scene contains rocks and multiple straps where
some of the straps contain holes. As shown in Figure 9, the
proposed semantic segmentation model could classify small
holes inside the straps as rocks, and this will help also for
drilling holes in straps safely. Figure 10 shows the ability of
the proposed model to distinguish power cables and straps
as non-rock regions. Figure 11 shows testing the proposed
deep learning semantic segmentation model on accumu-
lated images collected by an event camera. Even though the
deep learning model has not trained on images from event
cameras, it could provide acceptable results when it is tested
on accumulated event images. As shown in Figure 11, the
model could distinguish straps and power cables as non-
rock. This means that the proposed model not only appli-
cable for new test images, but it is also applicable for testing
images that are different from those that have been used for
the training of the model.
CONCLUSION
In this research, the authors proposed an adaptive deep
learning based semantic segmentation model which is nec-
essary to automate roof bolting in underground mine. This
research addressed the harsh environment in underground
Figure 6. Training and validation learning curves of the
losses for the proposed semantic segmentation model. The
lossesare decreasing and smooth after seven epochs.
Figure 7. Performance of the proposed deep learning based
semantic segmentation model in terms of accuracy during
the training and validation
Figure 8. Confusion matrix of the proposed semantic
segmentation model tested on unseen color images
• v1 score =96.36%
Comparing the learning accuracy in Figure 7 with the
testing accuracy in Figure 8, the proposed does not suf-
fer from overfitting or under-fitting which means that
the proposed model is robust and can be used in differ-
ent underground mines. The model succeeded to segment
color images that have been captured by a traditional cam-
era and classify the pixels that belong to rocks from pixels
that belong to straps. In addition, the model could locate
the holes in the straps which belong to rock.
Figure 9 shows the testing of the model for the scenario
where the scene contains rocks and multiple straps where
some of the straps contain holes. As shown in Figure 9, the
proposed semantic segmentation model could classify small
holes inside the straps as rocks, and this will help also for
drilling holes in straps safely. Figure 10 shows the ability of
the proposed model to distinguish power cables and straps
as non-rock regions. Figure 11 shows testing the proposed
deep learning semantic segmentation model on accumu-
lated images collected by an event camera. Even though the
deep learning model has not trained on images from event
cameras, it could provide acceptable results when it is tested
on accumulated event images. As shown in Figure 11, the
model could distinguish straps and power cables as non-
rock. This means that the proposed model not only appli-
cable for new test images, but it is also applicable for testing
images that are different from those that have been used for
the training of the model.
CONCLUSION
In this research, the authors proposed an adaptive deep
learning based semantic segmentation model which is nec-
essary to automate roof bolting in underground mine. This
research addressed the harsh environment in underground
Figure 6. Training and validation learning curves of the
losses for the proposed semantic segmentation model. The
lossesare decreasing and smooth after seven epochs.
Figure 7. Performance of the proposed deep learning based
semantic segmentation model in terms of accuracy during
the training and validation
Figure 8. Confusion matrix of the proposed semantic
segmentation model tested on unseen color images