Accurate potato surface defect segmentation is important for automated produce grading and post-harvest quality inspection, but remains challenging because defect regions are sparse and healthy-skin appearance varies substantially across individual tubers. This paper proposes PotatoRelSeg, a lightweight semantic segmentation framework based on intra-tuber relative representation learning. Instead of relying solely on absolute visual features or a shared class-level prototype, the framework estimates a sample-specific healthy-skin reference from coarse predictions and an image-derived foreground mask without requiring ground-truth annotations at inference. Pixel features are then represented jointly by their original descriptors, feature differences, cosine similarity, and Euclidean distance relative to the healthy reference and are refined through a compact depthwise-separable residual module. Healthy-reference alignment and local healthy-ring contrast further provide training supervision for the image-level reference and local defect–healthy relationships. Experiments on the predefined Diamant Potato Dataset split show that PotatoRelSeg achieves the strongest overall and defect-oriented segmentation performance among the evaluated methods. Sequential ablations further support the contribution of the proposed components, while efficiency analysis shows that the framework retains a compact model structure with a moderate computational overhead. These results suggest that intra-tuber reference conditioning provides a useful domain-specific inductive bias for potato surface defect segmentation.
The potato ranks as the world’s third most significant food crop, vital for human consumption and global food security, trailing only rice and wheat. India is a major global player, ranking as the second-largest producer worldwide. Processing potatoes is crucial globally, accounting for 60–70
Conservation agriculture may improve soil health, but its effects on wireworms and cutworms in potato systems remain unclear. A factorial experiment tested three rotations (R1: barley–potato, R2: faba bean–potato, R3: continuous potato), three tillage systems (T1: light tillage using a disc harrow, T2: chisel plowing, T3: moldboard plowing), and three weed management systems (W1: herbicide, W2: manual weeding, W3: no control). Agriotes obscurus, A. ustulatus, Agrotis ipsilon, and A. segetum were monitored using bait and pitfall traps, and crop damage and establishment were assessed over two seasons (2023–2024 and 2025–2026). Crop rotation had no significant effect on pest abundance, although total density increased from 72.0 larvae (barley–potato) to 92.5 larvae (continuous potato). Tillage significantly reduced wireworms, with moldboard plowing lowering A. obscurus and A. ustulatus to 4.0 and 2.7 larvae versus 15.9 and 14.1 under light tillage (T1), resulting in lower total pests (52.3 vs 102.7 larvae). Weed management also reduced cutworms, with herbicide plots showing only 3.9 A. ipsilon and 2.6 A. segetum compared with > 13 larvae per species in other treatments. T3R1W1 showed the best performance with 93.3–94.7
In general, the collaboration between technology and agriculture is getting stronger. Agriculture covers many crops, including vegetables and fruit types which are grown under diverse conditions. Climate and soil type are major factors during the growing process of potato. Another factor, the tuber itself, also plays a crucial role in determining characteristics of potato growth. It is therefore beneficial to distinguish different types of tubers, especially in the supply chain. Closer inspection of tubers could serve as an effective method for analyzing the elements and minerals within the potato once tubers have been correctly separated. Using machine learning in this field significantly contributes to obtaining more advanced conclusions. The Visual Geometry Group made the VGG-19 model, which was used to sort different cultivars of potatoes. Transfer learning was used by stopping the last layer of the neural network. Going further, five types of meta-heuristic methods (GA, PSO, ABC, ALO, and GWO) were used to get the best hyperparameters that yielded better results than without using meta-heuristics. Artificial Bee Colony (ABC) surpasses the other methods as well as the standard approach by reaching 90
Potato is a valuable crop both nutritionally and economically. However, its productivity is adversely affected by foliar diseases. Accurate and timely disease detection is critical. Deep learning models have achieved significant progress; however, limited work exists on specialized datasets. Most existing frameworks are trained on one or two deep learning models and use standard metrics evaluation, lacking multi-criteria decision-making for comprehensive model evaluation. To bridge the gaps in the available research, this study offers three key contributions. The creation of a well-labeled, balanced dataset consists of 5244 potato leaf images, representing six classes. The images were captured under different environmental conditions to accelerate model generalization. The evaluation of seven fine-tuned pre-trained convolutional neural networks (CNNs) on the benchmark dataset. The application of the evaluation based on distance from average solution (EDAS) method to multi-criteria model ranking, both overall and per-disease class, enables effective model selection. Experimental results demonstrate that InceptionV3 outperforms other models with a classification accuracy of 90.2
