Fruit detection in real agricultural environments has long been one of the most stubborn problems in computer vision, and a new study from researchers at the University of Lincoln and the University of Warwick now offers a fresh technical answer. In work published open access in Multimedia Tools and Applications, Dan Dai, Junfeng Gao, Elizabeth Sklar and Simon Parsons introduce ADNet_TL, an Adjustable Anchor Box Detection Network with Transfer Learning that is specifically designed to excel when training data is scarce. The team reports that the framework surpasses both their own baseline detector without transfer learning and the classical Single Shot MultiBox Detector, with gains of up to 14 percent in mean Average Precision across strawberry, tomato and multi-fruit image sets. The achievement matters because the promise of agricultural robotics, from autonomous picking to yield forecasting, often stalls at exactly this point: the models work well with vast labeled datasets but collapse when only a few dozen annotated images are available, which is the norm in real farming conditions.
The core difficulty the researchers confronted is threefold. Training data in agriculture is limited because photographing and annotating crops is slow and expensive, often requiring human labor in polytunnels and orchards. Fruit appearance varies dramatically across growth stages, so a tomato early in ripening looks little like the mature fruit a harvester ultimately needs to find. And occlusion is pervasive, since leaves, stems and neighboring fruit routinely hide much of the visual information a detector would like to see. Conventional detectors, whether anchor-based architectures such as Faster R-CNN and SSD or anchor-free approaches that predict object centers per pixel, typically assume fixed anchor configurations tuned on generic benchmarks and demand substantial labeled data. In unstructured agricultural scenes, those assumptions break down precisely where they are needed most.
At the heart of the new framework is the Fruit Discriminative Network, or FDN, a clever piece of weakly supervised engineering that sidesteps the need for exhaustive bounding-box annotation. The FDN is built on a VGG-16 backbone pretrained on ImageNet and performs only a simple binary classification: does an image contain fruit or not? Training images are drawn from two pools, a Without Fruit set of plants, leaves and backgrounds gathered from web searches, and a Detected Fruit set containing strawberries, tomatoes or a mixed collection of eleven fruits including apple, apricot, clementine, kaki, lemon, mango, orange, peach, pear, plum and strawberry. Because the task is binary, the network needs no pixel-level labels, and twenty epochs of training suffice.
The technical magic happens after classification, using Gradient-weighted Class Activation Mapping, or Grad-CAM. This technique computes the gradients flowing back to the final convolutional feature maps, weights the forward activations accordingly, and produces a spatial importance map showing which pixels most influenced the fruit-versus-no-fruit decision. Mathematically, the attention at each grid position is the sum over feature channels of the class-specific feedback gradients multiplied by the last-layer activations, passed through a rectified linear unit so that only positive contributions remain. The resulting heatmap, rendered across separate RGB channels, reveals exactly where the model believes fruit to be. Two kinds of Euclidean distance measurements are then extracted from this map: the distance between prominent highlighted regions and their non-important surroundings, and the distance between adjacent highlight areas. These distances provide a rough estimate of fruit size, which in turn encodes fruit type and growth state without any manual measurement.
Those size estimates feed directly into the second innovation, adaptive anchor box generation. In the standard SSD detector, six so-called anchor maps are defined by fixed lower and upper scale limits derived from hand-tuned hyperparameters, with a minimum scale computed from a base dimension of 300 pixels and ratios drawn from a predetermined set. The Lincoln team observed that such uniform settings are rarely optimal when fruit size distributions differ so widely between crops. Their algorithm inspects the attention map, identifies salient and non-salient pixel positions, and measures the Euclidean distances among them. If fruits tend to be gathered closely, as with clusters of grapes or densely packed tomatoes, distances within the salient set best reflect individual fruit size; if fruits are scattered, distances between salient and non-salient points are more informative. Based on this density assessment, the system selects one of two strategies. A frequency-based method suits datasets where most objects fall in a narrow size range: it takes the six most frequently occurring distances as the minimum scales of the six anchor maps and doubles them for the maximum scales. A linear-based method handles datasets with wide size variation, spacing the six anchor scales evenly between the minimum and maximum observed distances. Ablation experiments confirmed the intuition, with the frequency method performing better on the relatively uniform strawberry dataset and the linear method winning on tomato and multi-fruit data.
The third pillar is transfer learning, and here the study makes a contribution that goes beyond the usual practice by systematically exploring how the sizes of both source and target training sets affect performance. The team used the tomato dataset, collected from a garden and showing multiple growth stages, as the source domain, then fine-tuned on strawberry images from a commercial polytunnel in Lincoln, UK, and on the multi-fruit web collection. Fine-tuning froze the VGG convolutional layers and the localization layers while unfreezing the confidence layers, preserving initial weights there before continued training. Tomato training images were partitioned into subsets of 49, 99, 198 and 396 images, strawberry into subsets from 40 to 198 images, and multi-fruit into subsets from 40 to 336 images, each evaluated against fixed test sets of 50 strawberry and 111 multi-fruit images.
The results tell a nuanced story. Transferring knowledge from tomatoes to the multi-fruit dataset yielded the strongest gains, with the best combination of 336 target images and 396 source images reaching a mean Average Precision of 0.6409, nearly a 10 percent improvement over the non-transfer baseline. For a fixed tomato model trained on 198 images, raising multi-fruit training data from 40 to 120 images boosted mAP by 0.21, whereas the equivalent strawberry increase added only 0.03. To explain the asymmetry, the researchers visualized feature spaces extracted by VGG-16 using t-SNE, which converts similarities between data points into joint probabilities and minimizes the Kullback-Leibler divergence between high-dimensional data and a low-dimensional embedding. The two-dimensional projections showed that the tomato feature distribution largely covers that of the multi-fruit set, while strawberries, photographed against complex and cluttered backgrounds, occupy a much wider feature space, making them harder to transfer to.
Qualitatively, the detector’s outputs also proved more useful for actual farm robotics. Where the standard SSD often captured only part of a strawberry, offering insufficient information for a gripper, ADNet’s prediction boxes frequently included the stem, which is exactly what a picking mechanism needs to grasp. The adaptive anchors even located fruits whose color nearly matched the background, and handled clustered fruits markedly better. Interestingly, the Grad-CAM analysis uncovered a quirk: for tomatoes, the most discriminative regions were the stems rather than the fruit itself, because stems reliably co-occur with dense tomato clusters yet rarely appear in the fruitless background images. Rather than treating this as a failure, the researchers exploited it, using within-highlight distances to estimate the spacing between clustered tomatoes.
Efficiency figures round out the picture. ADNet needed only 3,000 training iterations to match the accuracy SSD achieved after 5,000 on the strawberry and tomato datasets, and loss curves fitted with twentieth-degree polynomials showed smoother, faster convergence on tomato and multi-fruit data. The FDN stage does add overhead, roughly 600 seconds for strawberries, 1,300 seconds for tomatoes and 235 seconds for the multi-fruit set, but the authors argue this one-time cost is offset by faster detection convergence and reduced annotation demands.
The work, supported by the EPSRC Centre for Doctoral Training in Agri-Food Robotics, positions ADNet_TL as a practical foundation for fruit forecasting and selective harvesting under genuinely unstructured conditions. The authors are candid about limitations, noting sensitivity to hyperparameter choices and residual domain-shift effects between source and target datasets. Their planned next steps include self-tuning hyperparameter techniques and adversarial and unsupervised domain adaptation, which could allow a single detector pretrained on one crop to generalize to many others with almost no local annotation at all.
Subject of Research: Robust fruit detection in plant phenotyping using adjustable anchor boxes and transfer learning for small datasets
Subject of Research: Technology and Engineering
Article Title: Fruit detection for small datasets via adjustable anchor boxes and transfer learning
Article References: Dai, D., Gao, J., Sklar, E., & Parsons, S. (2026). Fruit detection for small datasets via adjustable anchor boxes and transfer learning. Multimedia Tools and Applications, 85(8), Article 692. https://doi.org/10.1007/s11042-026-21246-1
Image Credits: AI Generated
DOI: 10.1007/s11042-026-21246-1
Keywords: Plant phenotyping, Fruit detection, Adjustable anchor boxes, Transfer learning, Grad-CAM, SSD, Weakly supervised learning, Object detection, Smart agriculture, Small datasets, mAP, Agricultural robotics
Cite Scienmag News
APA MLA Chicago
Denise Maddox. (September 9, 2026). Adjustable anchor boxes and transfer learning boost fruit detection on small datasets. Scienmag. https://scienmag.com/adjustable-anchor-boxes-and-transfer-learning-boost-fruit-detection-on-small-datasets/
Denise Maddox. “Adjustable anchor boxes and transfer learning boost fruit detection on small datasets.” Scienmag, 9 September 2026, https://scienmag.com/adjustable-anchor-boxes-and-transfer-learning-boost-fruit-detection-on-small-datasets/. Accessed 9 September 2026.
Denise Maddox. “Adjustable anchor boxes and transfer learning boost fruit detection on small datasets.” Scienmag. September 9, 2026. https://scienmag.com/adjustable-anchor-boxes-and-transfer-learning-boost-fruit-detection-on-small-datasets/
Copy citation Download RIS
Tags: adjustable anchor box detection networkagricultural robotics advancementsagricultural robotics and yield forecastingautonomous fruit pickingautonomous fruit picking technologychallenges in crop image annotationchallenges of limited annotated agricultural datacomputer vision in farmingdeep learning for agriculturedeep learning for crop monitoringfruit detection in agricultureimproving fruit detection accuracymean Average Precision improvementmulti-fruit image analysismulti-fruit image detectionsmall dataset fruit detectionstrawberry and tomato fruit recognitiontransfer learning for small datasetsyield forecasting with limited data
