ACADSTAFF UGM
| Title | : | Parking Lot Detection Using Deep Learning Mask Region-Convolutional Neural Network |
| Author | : |
Ika Candradewi, S.Si., M.Cs. (1) ARI FEBRIAN (2) |
| Date | : | 30 2020 |
| Keyword | : | parking slot status, classification, CNN, Mask R CNN parking slot status, classification, CNN, Mask R CNN |
| Abstract | : | The smart parking industry will continue to grow along with the increase in traffic congestion and the availability of insufficient parking. The results showed that as much as 30% of the traffic jams were caused by drivers who went around looking for parking lots in crowded city areas. This study aims to create a system to process the status of parking slots in car parking areas using the Mask Region-based Neural Network (Mask-RCNN) method in each parking area region and obtain the optimal model. The results of various configurations for parking slot classification by comparing the architecture, filter region proposal network (RPN), learning rate, and epoch of each model. In testing, the mean Average Precision (mAP) value of each model for all variations of the parking type with the best configuration is 97.3%. The architectural variation value is 81.7%, the variation in the dataset for parking type and the environment in the model with the highest value is 83.2%. The highest learning rate variation is 83.2%, and the filter region variation of the network proposal is 81.7%. |
| Group of Knowledge | : | Ilmu Komputer |
| Original Language | : | English |
| Level | : | Internasional |
| Status | : |
Under review
|