README.md

---
license: other
license_name: sla0044
license_link: >-
  https://github.com/STMicroelectronics/stm32ai-modelzoo/raw/refs/heads/main/human_activity_recognition/LICENSE.md
---
# ST_GMP HAR model

## **Use case** : `Human activity recognition`

# Model description

GMP is an acronym for Global Max Pooling. It is a convolutional neural network (CNN) based model that uses Global Max Pooling before feeding the data to the fully-connected (Dense) layer for performing the human activity recognition (HAR) task based on the accelerometer data. Prefix `st_` denotes it is a variation of the model built by STMicroelectronics. It uses the 3D raw data with gravity rotation and supression filter as preprocessing. This is a very light model with very small foot prints in terms of FLASH and RAM as well as computational requirements.

This network supports any input size greater than (3 x 3 x 1) but we recommend to use at least (24 x 3 x 1), i.e. a window length of 24 samples. In this folder we provide GMP models trained with two different window lenghts [24 and 48].

The only input required to the model is the input shape and the number of output classes.

In this folder you will find different copies of the GMP model pretrained on a public dataset ([WISDM](https://www.cis.fordham.edu/wisdm/dataset.php)) and a custom dataset collected by ST  (mobility_v1). 

## Network information (for WISDM at wl = 24)


| Network Information     |  Value          |
|:-----------------------:|:---------------:|
|  Framework              | TensorFlow      |
|  Params                 | 1,528           |


## Network inputs / outputs


For a frame of resolution of (wl x 3) and P classes

| Input Shape | Description |
| :----:| :-----------: |
| (1, wl, 3, 1) | Single ( wl x 3 x 1 ) matrix of accelerometer values, `wl` is window lenght, for 3 axes and 1 is channel in FLOAT32.|

| Output Shape | Description |
| :----:| :-----------: |
| (1, P) | Per-class confidence for P classes in FLOAT32|


## Recommended platforms


| Platform | Supported | Recommended |
|:--------:|:---------:|:-----------:|
| STM32L4  |    [x]    |      []     |
| STM32U5  |    [x]    |      [x]    |


# Performances

## Metrics

Measures are done with default STEdge AI Dev Cloud version 3.0.0 and for target board B-U585I-IOT02A. In addition the configuration were enabled input / output allocated option and `balanced` as optimization choice.

The inference time is reported is calculated on STM32 board **B-U585I-IOT02A** running at Frequency of **160 MHz**.


### Reference memory footprint based on WISDM dataset (see Accuracy for details on dataset)


| Model                                                                        | Format | Input Shape | Target Board  | Activation RAM (KiB) | Runtime RAM (KiB) | Weights Flash (KiB) | Code Flash (KiB) | Total RAM (KiB)   | Total Flash (KiB) | Inference Time (ms) | STEdge AI Core version  |
|:----------------------------------------------------------------------------:|:------:|:-----------:|:-------:|:--------------------:|:-----------------:|:-------------------:|:----------------:|:-----------------:|:-----------------:|:---------------------:|:---------------------:|
| [st_gpm_wl_24](ST_pretrainedmodel_public_dataset/WISDM/st_gmp_wl_24/st_gmp_wl_24.keras)  | FLOAT32| 24 x 3 x 1  | B-U585I-IOT02A | 4.25          | 0.28              | 5.70                | 6.08            |  4.53             | 11.78             | 4.29      | 3.0.0     |
| [st_gmp_wl_48](ST_pretrainedmodel_public_dataset/WISDM/st_gmp_wl_48/st_gmp_wl_48.keras)  | FLOAT32| 48 x 3 x 1  | B-U585I-IOT02A | 8.83          | 0.28              | 5.70                | 6.08            |  9.11            | 11.78             | 8.83     | 3.0.0     |



### Accuracy with mobility_v1 dataset


Dataset details: A custom dataset and not publically available, Number of classes: 5 [Stationary, Walking, Jogging, Biking, Vehicle]. **(We kept only 4, [Stationary, Walking, Jogging, Biking]) and removed Driving**, Number of input frames:  81,151 (for wl = 24), and 40,575 for (wl = 48).


| Model                                                                                          |  Format  | Resolution | Accuracy (%) |
|:----------------------------------------------------------------------------------------------:|:--------:|:----------:|:-------------:|
| [st_gmp_wl_24](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_gmp/ST_pretrainedmodel_custom_dataset/mobility_v1/st_gmp_wl_24/st_gmp_wl_24.keras)   | FLOAT32  | 24 x 3 x 1 | 93.93         |
| [st_gmp_wl_48](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_gmp/ST_pretrainedmodel_custom_dataset/mobility_v1/st_gmp_wl_48/st_gmp_wl_48.keras)   | FLOAT32  | 48 x 3 x 1 | 93.71         |


Confusion matrix for st_gmp_wl_24 with Float32 weights for mobility_v1 dataset is given below.

![plot](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_gmp/doc/img/mobility_v1_st_gmp_wl_24_confusion_matrix.png)

### Accuracy with WISDM dataset


Dataset details: [link](([WISDM](https://www.cis.fordham.edu/wisdm/dataset.php))) , License [CC BY 2.0](https://creativecommons.org/licenses/by/2.0/) , Quotation[[1]](#1) , Number of classes: 6 (we are **combining Upstairs and Downstairs into Stairs** and **Standing and Sitting into Stationary**), Number of samples: 45,579 (at wl = 24), and 22,880 (at wl = 48).

| Model                                                                                  |  Format  |  Resolution |  Accuracy (%)  |
|:--------------------------------------------------------------------------------------:|:--------:|:-----------:|:--------------:|
| [st_gmp_wl_24](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_gmp/ST_pretrainedmodel_public_dataset/WISDM/st_gmp_wl_24/st_gmp_wl_24.keras)    | FLOAT32  | 24 x 3 x 1  | 83.54          |
| [st_gmp_wl_48](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/st_gmp/ST_pretrainedmodel_public_dataset/WISDM/st_gmp_wl_48/st_gmp_wl_48.keras)    | FLOAT32  | 48 x 3 x 1  | 86.59          |


## Retraining and Integration in a simple example:

Please refer to the stm32ai-modelzoo-services GitHub [here](https://github.com/STMicroelectronics/stm32ai-modelzoo-services)



# References

<a id="1">[1]</a>
“WISDM : Human activity recognition datasets". [Online]. Available: "https://www.cis.fordham.edu/wisdm/dataset.php".