Release AI-ModelZoo-3.1.0:

- Support for STEdgeAI Core v2.1.0 (STM32Cube.AI v10.1.0).
- Application code for STM32N6 board is now directly available in the STM32 model zoo repository; eliminating the need for separate downloads.
- Support of On device evaluation and On device prediction on the STM32N6570-DK boards integrated in evaluation and prediction services.
- More models supported : Yolov11, LSTM model added in the Speech Enhancement, ST Yolo X variants.
- ClearML support.
- A few bug fixes and improvements like proper imports, OD metrics alignments.

Signed-off-by: khaoula boutiche <[email protected]>

Author: KBOUSTM

README.md

@@ -19,7 +19,11 @@ The performances on reference STM32 MCU, NPU and MPU are provided for float and
 
 ## What's new in releases :
 </details>
-<details open><summary><b>3.0:</b></summary>
+<details open><summary><b>3.1:</b></summary>
+
+* Included additional models support (yolo_v11, st_yolo_x variants)
+</details>
+<details><summary><b>3.0:</b></summary>
 
 * Included additional models compatible with the [STM32N6570-DK](https://www.st.com/en/evaluation-tools/stm32n6570-dk) board.
 * Expanded models in all use cases.

audio_event_detection/miniresnet/ST_pretrainedmodel_public_dataset/esc10/miniresnet_1stacks_64x50_tl/miniresnet_1stacks_64x50_tl_config.yaml

@@ -16,8 +16,8 @@ dataset:
   name: esc10
   class_names: ['dog', 'chainsaw', 'crackling_fire', 'helicopter', 'rain', 'crying_baby', 'clock_tick', 'sneezing', 'rooster', 'sea_waves']
   file_extension: '.wav'
-  training_audio_path: ..\datasets\ESC-50\audio # Mandatory
-  training_csv_path:   ..\datasets\ESC-50\meta\esc50.csv # Mandatory
+  training_audio_path: ./datasets/ESC-50/audio # Mandatory
+  training_csv_path:   ./datasets/ESC-50/meta/esc50.csv # Mandatory
 
   validation_audio_path: # Optional
   validation_csv_path: # Optional
@@ -145,7 +145,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -156,8 +156,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/miniresnet_1stack_esc10_for_real
+    dir: ./src/experiments_outputs/miniresnet_1stack_esc10_for_real

audio_event_detection/miniresnet/ST_pretrainedmodel_public_dataset/esc10/miniresnet_2stacks_64x50_tl/miniresnet_2stacks_64x50_tl_config.yaml

@@ -16,8 +16,8 @@ dataset:
   name: esc10
   class_names: ['dog', 'chainsaw', 'crackling_fire', 'helicopter', 'rain', 'crying_baby', 'clock_tick', 'sneezing', 'rooster', 'sea_waves']
   file_extension: '.wav'
-  training_audio_path: ..\datasets\ESC-50\audio # Mandatory
-  training_csv_path:   ..\datasets\ESC-50\meta\esc50.csv # Mandatory
+  training_audio_path: ./datasets/ESC-50/audio # Mandatory
+  training_csv_path:   ./datasets/ESC-50/meta/esc50.csv # Mandatory
 
   validation_audio_path: # Optional
   validation_csv_path: # Optional
@@ -146,7 +146,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -157,8 +157,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/miniresnet_2stack_esc10
+    dir: ./src/experiments_outputs/miniresnet_2stack_esc10

audio_event_detection/miniresnetv2/ST_pretrainedmodel_public_dataset/esc10/miniresnetv2_1stacks_64x50_tl/miniresnetv2_1stacks_64x50_tl_config.yaml

@@ -16,8 +16,8 @@ dataset:
   name: esc10
   class_names: ['dog', 'chainsaw', 'crackling_fire', 'helicopter', 'rain', 'crying_baby', 'clock_tick', 'sneezing', 'rooster', 'sea_waves']
   file_extension: '.wav'
-  training_audio_path: ..\datasets\ESC-50\audio # Mandatory
-  training_csv_path:   ..\datasets\ESC-50\meta\esc50.csv # Mandatory
+  training_audio_path: ./datasets/ESC-50/audio # Mandatory
+  training_csv_path:   ./datasets/ESC-50/meta/esc50.csv # Mandatory
 
   validation_audio_path: # Optional
   validation_csv_path: # Optional
@@ -145,7 +145,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -156,8 +156,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/miniresnetv2_1stack_esc10
+    dir: ./src/experiments_outputs/miniresnetv2_1stack_esc10

audio_event_detection/miniresnetv2/ST_pretrainedmodel_public_dataset/esc10/miniresnetv2_2stacks_64x50_tl/miniresnetv2_2stacks_64x50_tl_config.yaml

@@ -16,8 +16,8 @@ dataset:
   name: esc10
   class_names: ['dog', 'chainsaw', 'crackling_fire', 'helicopter', 'rain', 'crying_baby', 'clock_tick', 'sneezing', 'rooster', 'sea_waves']
   file_extension: '.wav'
-  training_audio_path: ..\datasets\ESC-50\audio # Mandatory
-  training_csv_path:   ..\datasets\ESC-50\meta\esc50.csv # Mandatory
+  training_audio_path: ./datasets/ESC-50/audio # Mandatory
+  training_csv_path:   ./datasets/ESC-50/meta/esc50.csv # Mandatory
 
   validation_audio_path: # Optional
   validation_csv_path: # Optional
@@ -145,7 +145,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -156,8 +156,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/miniresnetv2_2stack_esc10
+    dir: ./src/experiments_outputs/miniresnetv2_2stack_esc10

audio_event_detection/yamnet/ST_pretrainedmodel_public_dataset/esc10/yamnet_1024_64x96_tl/yamnet_1024_64x96_tl.yaml

@@ -1,6 +1,6 @@
 general:
   project_name: aed_project
-  model_path: /local/home/martinlu/model_zoo_services/audio_event_detection/src/experiments_outputs/2024_10_01_12_20_04/saved_models/best_model.h5
+  model_path:
   logs_dir: logs
   saved_models_dir: saved_models
   global_seed: 120
@@ -146,8 +146,8 @@ benchmarking:
   board: B-U585I-IOT02A 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/${now:%Y_%m_%d_%H_%M_%S}
+    dir: ./src/experiments_outputs/${now:%Y_%m_%d_%H_%M_%S}

audio_event_detection/yamnet/ST_pretrainedmodel_public_dataset/esc10/yamnet_256_64x96_tl/yamnet_256_64x96_tl_config.yaml

@@ -16,8 +16,8 @@ dataset:
   name: esc10
   class_names: ['dog', 'chainsaw', 'crackling_fire', 'helicopter', 'rain', 'crying_baby', 'clock_tick', 'sneezing', 'rooster', 'sea_waves']
   file_extension: '.wav'
-  training_audio_path: ..\datasets\ESC-50\audio # Mandatory
-  training_csv_path:   ..\datasets\ESC-50\meta\esc50.csv # Mandatory
+  training_audio_path: ./datasets/ESC-50/audio # Mandatory
+  training_csv_path:   ./datasets/ESC-50/meta/esc50.csv # Mandatory
 
   validation_audio_path: # Optional
   validation_csv_path: # Optional
@@ -143,7 +143,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -154,8 +154,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/yamnet_256_esc_10_second_run
+    dir: ./src/experiments_outputs/yamnet_256_esc_10_second_run

audio_event_detection/yamnet/ST_pretrainedmodel_public_dataset/fsd50k/yamnet_256_64x96_tl/with_unknown_class/yamnet_256_64x96_tl_config.yaml

@@ -44,12 +44,12 @@ dataset_specific:
   # Currently only supports fsd50k.
   # These parameters only need to be filled out IF the dataset name is set to 'fsd50K'
   fsd50k:
-    csv_folder: ../datasets/FSD50K/FSD50K.ground_truth
-    dev_audio_folder: ../datasets/FSD50K/FSD50K.dev_audio
-    eval_audio_folder: ../datasets/FSD50K/FSD50K.eval_audio
+    csv_folder: ./datasets/FSD50K/FSD50K.ground_truth
+    dev_audio_folder: ./datasets/FSD50K/FSD50K.dev_audio
+    eval_audio_folder: ./datasets/FSD50K/FSD50K.eval_audio
     # Change this next line to the ontology path on your machine. 
     # Download the ontology at https://github.com/audioset/ontology
-    audioset_ontology_path: preprocessing/dataset_utils/fsd50k/audioset_ontology.json 
+    audioset_ontology_path: ./src/preprocessing/dataset_utils/fsd50k/audioset_ontology.json 
     only_keep_monolabel: True
 
 preprocessing:
@@ -157,7 +157,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -168,8 +168,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/whatever
+    dir: ./src/experiments_outputs/whatever

audio_event_detection/yamnet/ST_pretrainedmodel_public_dataset/fsd50k/yamnet_256_64x96_tl/without_unknown_class/yamnet_256_64x96_tl_config.yaml

@@ -44,12 +44,12 @@ dataset_specific:
   # Currently only supports fsd50k.
   # These parameters only need to be filled out IF the dataset name is set to 'fsd50K'
   fsd50k:
-    csv_folder: ../datasets/FSD50K/FSD50K.ground_truth
-    dev_audio_folder: ../datasets/FSD50K/FSD50K.dev_audio
-    eval_audio_folder: ../datasets/FSD50K/FSD50K.eval_audio
+    csv_folder: ./datasets/FSD50K/FSD50K.ground_truth
+    dev_audio_folder: ./datasets/FSD50K/FSD50K.dev_audio
+    eval_audio_folder: ./datasets/FSD50K/FSD50K.eval_audio
     # Change this next line to the ontology path on your machine. 
     # Download the ontology at https://github.com/audioset/ontology
-    audioset_ontology_path: preprocessing/dataset_utils/fsd50k/audioset_ontology.json
+    audioset_ontology_path: ./src/preprocessing/dataset_utils/fsd50k/audioset_ontology.json
     only_keep_monolabel: True
 
 preprocessing:
@@ -143,7 +143,7 @@ quantization:
 # The prediction section is optional unless you set operation_mode to "prediction".
 # If you do so, the attribute test_files_path is mandatory.
 prediction:
-  test_files_path: ../test_files
+  test_files_path: ./datasets/test_files
 
 tools:
   stedgeai:
@@ -157,7 +157,7 @@ benchmarking:
   board: B-U585I-IOT02A
 
 deployment:
-  c_project_path: ../../application_code/sensing_free_rtos
+  c_project_path: ../application_code/sensing_free_rtos
   IDE: GCC
   verbosity: 1
   hardware_setup:
@@ -168,8 +168,8 @@ deployment:
 
 
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 
 hydra:
   run:
-    dir: ./experiments_outputs/whatever
+    dir: ./src/experiments_outputs/whatever

hand_posture/CNN2D_ST_HandPosture/ST_pretrainedmodel_custom_dataset/ST_VL53L5CX_handposture_dataset/CNN2D_ST_HandPosture_8classes/CNN2D_ST_HandPosture_8classes_config.yaml

@@ -19,7 +19,7 @@ dataset:
   - Love
   - BreakTime
   - CrossHands
-  training_path: ../datasets/ST_VL53L5CX_handposture_dataset
+  training_path: ./datasets/ST_VL53L5CX_handposture_dataset
   validation_path: null
   validation_split: null
   test_path: null
@@ -62,14 +62,14 @@ tools:
 benchmarking:
   board: NUCLEO-F401RE
 deployment:
-  c_project_path: ../../application_code/hand_posture/
+  c_project_path: ../application_code/hand_posture/
   IDE: GCC
   verbosity: 1
   hardware_setup:
     serie: STM32F4
     board: NUCLEO-F401RE
 mlflow:
-  uri: ./experiments_outputs/mlruns
+  uri: ./src/experiments_outputs/mlruns
 hydra:
   run:
-    dir: ./experiments_outputs/${now:%Y_%m_%d_%H_%M_%S}
+    dir: ./src/experiments_outputs/${now:%Y_%m_%d_%H_%M_%S}