BoltNet: An Ultra-Lightweight Convolutional Network for On-Device Plant Species Identification

2026-08-12

AI Summary

This work addresses the challenge of achieving high-accuracy, fine-grained plant species identification on resource-constrained mobile devices by proposing BoltNet, an ultra-lightweight fully convolutional network. BoltNet incorporates a spatial redistribution bottleneck and a logit pre-sampling mechanism to effectively balance classification accuracy and model size in large-scale species recognition tasks. The study introduces Accuracy-Compression Tradeoff as a key evaluation metric to emphasize real-world hardware deployment efficiency. With only 341K parameters (1.37 MB), BoltNet achieves a state-of-the-art F1-score of 0.682 on the Pl@ntNet300K benchmark among models under 2 MB. It further demonstrates superior energy efficiency across diverse edge platforms, including Raspberry Pi 5, Jetson Orin Nano, and Hailo-8.

Abstract

Automated plant species identification from citizen-science imagery is an established, demanding fine-grained recognition problem: large taxonomic label spaces, visually similar species, and long-tailed observations require real model capacity, while field use constrains memory, latency, and power. Model size is only part of the deployment cost: intermediate activations held in memory during inference and platformdependent execution behavior matter too, so compact recognition must be assessed on target hardware rather than through complexity metrics alone. We present BoltNet, an ultra-lightweight fully convolutional architecture combining a Spatial Redistribution Bottleneck and Logit PreSampling to improve the tradeoff between predictive performance and model size in high-cardinality classification, and report the AccuracyCompression Tradeoff as a complementary diagnostic. On Pl@ntNet300K, BoltNet reaches 0.682 F1-score with 341K parameters (1.37 MB), the highest F1-score among evaluated models below 2 MB and close to substantially larger convolutional backbones. Model-only measurements on a Raspberry Pi 5, Jetson Orin Nano, and Hailo-8 characterize execution across CPU, GPU, and NPU platforms, where BoltNet is the most consistently efficient model, with the best FPS/W on the GPU and NPU and second-best on the CPU. Results on AIDERv2 and CLRS provide secondary evidence of transfer across environmental image-classification tasks. Code available at: https://codeberg.org/danielrossi/BoltNet

Problem

• plant species identification
• fine-grained recognition
• on-device inference
• model compression
• resource-constrained deployment

Innovation

• BoltNet
• ultra-lightweight CNN
• Spatial Redistribution Bottleneck
• Logit PreSampling
• on-device inference