Release AI-ModelZoo-4.0.0

README.md

@@ -4,13 +4,13 @@ license_name: sla0044
 license_link: >-
   https://github.com/STMicroelectronics/stm32ai-modelzoo/raw/refs/heads/main/human_activity_recognition/LICENSE.md
 ---
-# GMP HAR model
+# 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. 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.
+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].
 
@@ -54,7 +54,7 @@ For a frame of resolution of (wl x 3) and P classes
 
 ## Metrics
 
-Measures are done with default STM32Cube.AI Dev Cloud version 10.2.0 and for target board B-U585I-IOT02A. In addition the configuration were enabled input / output allocated option and `balanced` as optimization choice.
+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**.
 
@@ -62,10 +62,10 @@ The inference time is reported is calculated on STM32 board **B-U585I-IOT02A** r
 ### 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) | STM32Cube.AI version  |
+| 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  |
 |:----------------------------------------------------------------------------:|:------:|:-----------:|:-------:|:--------------------:|:-----------------:|:-------------------:|:----------------:|:-----------------:|:-----------------:|:---------------------:|:---------------------:|
-| [GMP wl 24](ST_pretrainedmodel_public_dataset/WISDM/gmp_wl_24/gmp_wl_24.h5)  | FLOAT32| 24 x 3 x 1  | B-U585I-IOT02A | 4.25          | 2.08              | 5.70                | 12.29            |  6.33             | 18.96             | 4.42      | 10.2.0     |
-| [GMP wl 48](ST_pretrainedmodel_public_dataset/WISDM/gmp_wl_48/gmp_wl_48.h5)  | FLOAT32| 48 x 3 x 1  | B-U585I-IOT02A | 8.83          | 2.08              | 5.70                | 12.29            |  10.91            | 18.96             | 10.64     | 10.2.0     |
+| [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     |
 
 
 
@@ -77,13 +77,13 @@ Dataset details: A custom dataset and not publically available, Number of classe
 
 | Model                                                                                          |  Format  | Resolution | Accuracy (%) |
 |:----------------------------------------------------------------------------------------------:|:--------:|:----------:|:-------------:|
-| [GMP wl 24](./ST_pretrainedmodel_custom_dataset/mobility_v1/gmp_wl_24/gmp_wl_24.h5)            | FLOAT32  | 24 x 3 x 1 | 94.08         |
-| [GMP wl 48](./ST_pretrainedmodel_custom_dataset/mobility_v1/gmp_wl_48/gmp_wl_48.h5)            | FLOAT32  | 48 x 3 x 1 | 93.84         |
+| [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 GMP wl 24 with Float32 weights for mobility_v1 dataset is given below.
+Confusion matrix for st_gmp_wl_24 with Float32 weights for mobility_v1 dataset is given below.
 
-![plot](./doc/img/mobility_v1_gmp_wl_24_confusion_matrix.png)
+![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
 
@@ -92,8 +92,8 @@ Dataset details: [link](([WISDM](https://www.cis.fordham.edu/wisdm/dataset.php))
 
 | Model                                                                                  |  Format  |  Resolution |  Accuracy (%)  |
 |:--------------------------------------------------------------------------------------:|:--------:|:-----------:|:--------------:|
-| [GMP wl 24](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/gmp/ST_pretrainedmodel_public_dataset/WISDM/gmp_wl_24/gmp_wl_24.h5)            | FLOAT32  | 24 x 3 x 1  | 84.49          |
-| [GMP wl 48](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/human_activity_recognition/gmp/ST_pretrainedmodel_public_dataset/WISDM/gmp_wl_48/gmp_wl_48.h5)            | FLOAT32  | 48 x 3 x 1  | 87.05          |
+| [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: