Pseudo-Label Refinement for Robust Wheat Head Segmentation via Two-Stage Hybrid Training

2025/12/07 by Jiahao Jiang, Jiang, Jiahao, Zhengguo Yang +5
Agricultural and Biological Sciences · Computer Science · #Smart Agriculture and AI #Advanced Neural Network Applications #Machine Learning and Data Classification

paper · pdf · doi:10.48550/arxiv.2512.11874

Abstract

This extended abstract details our solution for the Global Wheat Full Semantic Segmentation Competition. We developed a systematic self-training framework. This framework combines a two-stage hybrid training strategy with extensive data augmentation. Our core model is SegFormer with a Mix Transformer (MiT-B4) backbone. We employ an iterative teacher-student loop. This loop progressively refines model accuracy. It also maximizes data utilization. Our method achieved competitive performance. This was evident on both the Development and Testing Phase datasets.

Citations

• Semi-Self-Supervised Domain Adaptation: Developing Deep Learning Models with Limited Annotated Data for Wheat Head Segmentation
• Micrograph segmentations for DDEVD
• Masked-attention Mask Transformer for Universal Image Segmentation
• SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers
• U-Net: Convolutional Networks for Biomedical Image Segmentation
• Distilling the Knowledge in a Neural Network
• Fully Convolutional Networks for Semantic Segmentation

Related

• Modified CycleGAN for the synthesization of samples for wheat head segmentation
• First Place Solution to the MLCAS 2025 GWFSS Challenge: The Devil is in the Detail and Minority
• Global Wheat Challenge 2020: Analysis of the competition design and winning models
• SynthSet: Generative Diffusion Model for Semantic Segmentation in Precision Agriculture
• Semi-Self-Supervised Domain Adaptation: Developing Deep Learning Models with Limited Annotated Data for Wheat Head Segmentation