The Effect of Activity Granularity on A Kitchen Activity Recognition System
Bibliographic record
Abstract
Kitchen activity recognition enables tailored advertising based on users' cooking habits and preferences, enhancing marketing relevance. Like any other human activity, the granularity of the kitchen activity that a sensor aims to recognize may affect the sensor's performance. This study aims to investigate the effect of activity granularity on the performance of a Wi-Fi-based human activity recognition (HAR) method in a kitchen context. For this purpose, first, we utilized an authentic kitchen with an ESP32 microcontroller as a WiFi transceiver and an iPhone 12 mini as a WiFi receiver. Then, we asked one user to perform three coarse-grained kitchen activities: filling an electric kettle with water and turning it on, stir-frying cubed potatoes, and taking several cans from the fridge and putting them on a cabinet. We gathered the Channel State Information (CSI) and Received Signal Strength Indicator (RSSI) data from WiFi packets. Then, we designed and implemented a device-free HAR system by a combination of CSI and RSSI at the feature level and utilizing a 5-fold cross-validation to assess the performance of a voting-based hybrid classification method including support vector machine (SVM), linear discriminant analysis (LDA), and Gaussian Naïve Bayes (GNB). The proposed system achieved an average accuracy of 92.27% in the recognition of coarse-grained activities. We then repeated the same experiment for gathering the same data for three fine-grained activities: chopping, slicing, and French-fries cutting. The same HAR system achieved an average accuracy of 51.53%. The results indicate that a device-free HAR method that achieves a high recognition accuracy for coarse-grained activities cannot achieve a similarly high accuracy for fine-grained activities.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".