| Title |
Shape based Plant Growth Prediction via Hierarchical Auto-encoder |
| Authors |
김태현(Tae-Hyeon Kim) ; 이상호(Sang-Ho Lee) ; 임은정(Eun-Jeong Lim) ; 오명민(Myung-Min Oh) ; 김종옥(Jong-Ok Kim) |
| DOI |
https://doi.org/10.5573/ieie.2022.59.7.69 |
| Keywords |
Image prediction; Spatial transformer network; Vision transformer; Plant growth; Shape domain |
| Abstract |
Plant growth prediction is significantly challenging because the growth rate varies depending on environmental factors. It is an essential task for efficient cultivation in controlled environments such as plant factory. In this paper, we propose a novel deep learning network to predict the future plant image from a current one. In particular, our focus is placed on the estimation of leaf shape in a plant because the amount of plant growth is commonly quantified by the leaf area. In the shape prediction sub-network, the future plant shape is first estimated using a hierarchical auto-encoder. After estimating the future plant shape, the RGB information is restored. After extracting feature maps using Vision Transformers, the current RGB plant image is fused with the predicted plant shape to generate a future RGB plant image. Experimental results show that the proposed network is robust to dydamic leaf movement and growth of various plants and can accurately predict the shape of the future plant image. |