# SmellNet: A Large-scale Dataset for Real-world Smell Recognition Source: [SmellNet: A Large-scale Dataset for Real-world Smell Recognition]() Collection index: [llms.txt]() Data updated: 2026-04-16T05:00:04.684Z Authors: Dewei Feng, Wei Dai, Carol Li, Alistair Pernigo, Yunge Wen, Paul Pu Liang Published: 2025-05-30T21:15:25.000Z ## Abstract The ability of AI to sense and identify various substances based on their smell alone can have profound impacts on allergen detection (e.g. smelling gluten or peanuts in a cake), monitoring the manufacturing process, and sensing hormones that indicate emotional states, stress levels, and diseases. Despite these broad impacts, there are few standardized datasets, and therefore little progress, for training and evaluating AI systems' ability to \`smell' in the real-world. In this paper, we use small gas and chemical sensors to create SmellNet, a comparatively large dataset for sensor-based machine olfaction that digitizes a diverse range of smells in the natural world. SmellNet contains about 828,000 time-series data points across 50 substances, spanning nuts, spices, herbs, fruits, and vegetables, and 43 mixtures among them with fixed ingredient volumetric ratios, with 68 hours of data collected. Using SmellNet, we developed ScentFormer, a Transformer-based architecture combining temporal differencing and sliding-window augmentation for smell data. For the SmellNet-Base classification tasks, ScentFormer achieves 63.3% Top-1 accuracy with GC-MS supervision, and for the SmellNet-Mixture distribution prediction tasks, ScentFormer achieves 50.2% Top-1@0.1 on the test-seen split. ScentFormer's ability to generalize across conditions and capture transient chemical dynamics demonstrates the promise of temporal modeling in sensor-based olfactory AI. SmellNet and ScentFormer lay the groundwork for sensor-based olfactory applications across healthcare, food and beverage, environmental monitoring, manufacturing, and entertainment. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Single-ingredient odor classification on SMELLNET BASE](): Single-ingredient odor classification; SMELLNET BASE; Accuracy@1: 63.3. - [Distribution Prediction on SMELLNET-MIXTURE BASE variant (seen combinations)](): Distribution prediction; SMELLNET-MIXTURE BASE variant (seen combinations); MAE: 0.0395. - [Distribution Prediction on SMELLNET-MIXTURE (unseen combinations)](): Distribution prediction; SMELLNET-MIXTURE (unseen combinations); Top-1 Accuracy @ 0.1: 16. - [Mixture proportion prediction on SMELLNET MIXTURE (seen)](): Mixture proportion prediction; SMELLNET MIXTURE (seen); KL Divergence: 0.386. - [Smell-to-Smell Retrieval on SmellNet (test)](): Smell-to-Smell Retrieval; SmellNet (test); R@1: 52.54. - [Mixture Proportion Prediction on SMELLNET-MIXTURE (unseen)](): Mixture proportion prediction; SMELLNET-MIXTURE (unseen); KL Divergence: 2.454. - [Family-wise Classification on SmellNet V (unseen ingredients)](): Family-wise Classification; SmellNet V (unseen ingredients); Accuracy: 46.58. - [Smell Classification on SmellNet smell-only (test)](): Smell Classification; SmellNet smell-only (test); Accuracy: 57.9. ## Original sources and implementation [Original paper]() [PDF]()