Hugging Face's logo

STMicroelectronics
/
movenet

Keypoint Detection
Model card Files
xet
Community
FBAGSTM commited on
Commit
72e5e32
·
verified ·
1 Parent(s): c64ec8f

Release AI-ModelZoo-4.0.0

Browse files
Files changed (1) hide show
  1. README.md +55 -38
README.md CHANGED
@@ -72,60 +72,77 @@ With an image resolution of NxM with K keypoints to detect :
72
  Measures are done with default STM32Cube.AI configuration with enabled input / output allocated option.
73
 
74
  ### Reference **NPU** memory footprint based on COCO Person dataset (see Accuracy for details on dataset)
75
- |Model | Dataset | Format | Resolution | Series | Internal RAM (KiB)| External RAM (KiB) | Weights Flash (KiB) | STM32Cube.AI version | STEdgeAI Core version |
76
- |----------|------------------|--------|-------------|------------------|------------------|---------------------|-------|----------------------|-------------------------|
77
- | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/custom_dataset_person_13kpts/st_movenet_lightning_heatmaps_192/st_movenet_lightning_heatmaps_192_int8_pc.tflite) | COCO-Person | Int8 | 192x192x3 | STM32N6 | 1674 | 0.0 | 3036.17 | 10.2.0 | 2.2.0 |
78
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_192/movenet_lightning_heatmaps_192_int8_pc.tflite) | COCO-Person | Int8 | 192x192x3 | STM32N6 | 1674 | 0.0 | 3036.41 | 10.2.0 | 2.2.0 |
79
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_224/movenet_lightning_heatmaps_224_int8_pc.tflite) | COCO-Person | Int8 | 224x224x3 | STM32N6 | 2058 | 0.0 | 3088.56 | 10.2.0 | 2.2.0 |
80
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_256/movenet_lightning_heatmaps_256_int8_pc.tflite) | COCO-Person | Int8 | 256x256x3 | STM32N6 | 2360 | 0.0 | 3141.36 | 10.2.0 | 2.2.0 |
81
 
82
 
83
  ### Reference **NPU** inference time based on COCO Person dataset (see Accuracy for details on dataset)
84
- | Model | Dataset | Format | Resolution | Board | Execution Engine | Inference time (ms) | Inf / sec | STM32Cube.AI version | STEdgeAI Core version |
85
- |--------|------------------|--------|-------------|------------------|------------------|---------------------|-------|----------------------|-------------------------|
86
- | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/custom_dataset_person_13kpts/st_movenet_lightning_heatmaps_192/st_movenet_lightning_heatmaps_192_int8_pc.tflite) | COCO-Person | Int8 | 192x192x3 | STM32N6570-DK | NPU/MCU | 18.44 | 54.23 | 10.2.0 | 2.2.0 |
87
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_192/movenet_lightning_heatmaps_192_int8_pc.tflite) | COCO-Person | Int8 | 192x192x3 | STM32N6570-DK | NPU/MCU | 18.49 | 54.08 | 10.2.0 | 2.2.0 |
88
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_224/movenet_lightning_heatmaps_224_int8_pc.tflite) | COCO-Person | Int8 | 224x224x3 | STM32N6570-DK | NPU/MCU | 22.33 | 44.78 | 10.2.0 | 2.2.0 |
89
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_256/movenet_lightning_heatmaps_256_int8_pc.tflite) | COCO-Person | Int8 | 256x256x3 | STM32N6570-DK | NPU/MCU | 27.01 | 37.03 | 10.2.0 | 2.2.0 |
90
 
91
 
92
  ### Reference **MPU** inference time based on COCO Person dataset (see Accuracy for details on dataset)
93
- | Model | Format | Resolution | Quantization | Board | Execution Engine | Frequency | Inference time (ms) | %NPU | %GPU | %CPU | X-LINUX-AI version | Framework |
94
- |--------------------------------------------------------------------------------------------------------------------------------------------------------------------|--------|------------|---------------|-------------------|------------------|-----------|---------------------|-------|-------|------|--------------------|-----------------------|
95
- | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/custom_dataset_person_13kpts/st_movenet_lightning_heatmaps_192/st_movenet_lightning_heatmaps_192_int8_pc.tflite) | Int8 | 192x192x3 | per-channel** | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 55.74 ms | 3.56 | 96.44 |0 | v6.1.0 | OpenVX |
96
- | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/custom_dataset_person_13kpts/st_movenet_lightning_heatmaps_192/st_movenet_lightning_heatmaps_192_int8_pt.tflite) | Int8 | 192x192x3 | per-tensor | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 7.82 ms | 84.47 | 14.53 |0 | v6.1.0 | OpenVX |
97
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_192/movenet_lightning_heatmaps_192_int8_pc.tflite) | Int8 | 192x192x3 | per-channel** | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 55.74 ms | 3.58 | 96.42 |0 | v6.1.0 | OpenVX |
98
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_192/movenet_lightning_heatmaps_192_int8_pt.tflite) | Int8 | 192x192x3 | per-tensor | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 8.17 ms | 86.37 | 13.63 |0 | v6.1.0 | OpenVX |
99
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_224/movenet_lightning_heatmaps_224_int8_pc.tflite) | Int8 | 224x224x3 | per-channel** | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 79.14 ms | 3.17 | 96.83 |0 | v6.1.0 | OpenVX |
100
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_224/movenet_lightning_heatmaps_224_int8_pt.tflite) | Int8 | 224x224x3 | per-tensor | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 11.26 ms | 86.76 | 13.24 |0 | v6.1.0 | OpenVX |
101
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_256/movenet_lightning_heatmaps_256_int8_pc.tflite) | Int8 | 256x256x3 | per-channel** | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 68.02 ms | 3.54 | 96.46 |0 | v6.1.0 | OpenVX |
102
- | [MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_256/movenet_lightning_heatmaps_256_int8_pc.tflite) | Int8 | 256x256x3 | per-tensor | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 12.90 ms | 86.33 | 13.67 |0 | v6.1.0 | OpenVX |
103
- | [MoveNet Lightning](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_192/movenet_singlepose_lightning_192_int8.tflite) | Int8 | 192x192x3 | per-channel** | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 64.2 ms | 6.83 | 93.17 |0 | v6.1.0 | OpenVX
104
- | [MoveNet Thunder](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_thunder_256/movenet_singlepose_thunder_256_int8.tflite) | Int8 | 256x256x3 | per-channel** | STM32MP257F-DK2 | NPU/GPU | 800 MHz | 183.82 ms | 3.43 | 96.57 |0 | v6.1.0 | OpenVX |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
105
 
106
  ** **To get the most out of MP25 NPU hardware acceleration, please use per-tensor quantization**
107
 
 
 
108
  ### OKS on COCO Person dataset
109
 
110
 
111
  Dataset details: [link](https://cocodataset.org/#download) , License [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/legalcode) , Quotation[[1]](#1) , Number of classes: 80, Number of images: 118,287
112
 
113
 
114
- | Model | Format | Resolution | OKS |
115
- |-------|--------|------------|----------------|
116
- | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/custom_dataset_person_13kpts/st_movenet_lightning_heatmaps_192/st_movenet_lightning_heatmaps_192_int8_pc.tflite) | Int8 | 192x192x3 | *52.1 % |
117
- | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/custom_dataset_person_13kpts/st_movenet_lightning_heatmaps_192/st_movenet_lightning_heatmaps_192_int8_pt.tflite) | Int8 | 192x192x3 | *39.31 % |
118
- | [MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_192/movenet_lightning_heatmaps_192_int8_pc.tflite) | Int8 | 192x192x3 | 54.01 % |
119
- | [MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_192/movenet_lightning_heatmaps_192_int8_pt.tflite) | Int8 | 192x192x3 | 48.49 % |
120
- | [MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_224/movenet_lightning_heatmaps_224_int8_pc.tflite) | Int8 | 224x224x3 | 57.07 % |
121
- | [MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_224/movenet_lightning_heatmaps_224_int8_pt.tflite) | Int8 | 224x224x3 | 50.93 % |
122
- | [MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_256/movenet_lightning_heatmaps_256_int8_pc.tflite) | Int8 | 256x256x3 | 58.58 % |
123
- | [MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_heatmaps_256/movenet_lightning_heatmaps_256_int8_pt.tflite) | Int8 | 256x256x3 | 52.86 % |
124
- | [MoveNet Lightning](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_192/movenet_singlepose_lightning_192_int8.tflite) | Int8 | 192x192x3 | 54.12% |
125
- | [MoveNet Thunder](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_thunder_256/movenet_singlepose_thunder_256_int8.tflite) | Int8 | 256x256x3 | 64.43% |
126
-
 
 
 
127
 
128
- \* keypoints = 13
129
 
130
  ## Integration in a simple example and other services support:
131
 
 
72
  Measures are done with default STM32Cube.AI configuration with enabled input / output allocated option.
73
 
74
  ### Reference **NPU** memory footprint based on COCO Person dataset (see Accuracy for details on dataset)
75
+ |Model | Dataset | Format | Resolution | Series | Internal RAM (KiB)| External RAM (KiB) | Weights Flash (KiB) | STEdgeAI Core version |
76
+ |----------|------------------|--------|-------------|------------------|------------------|---------------------|-------|-------------------------|
77
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | COCO-Person | Int8 | 192x192x3 | STM32N6 | 914.88 | 0.0 | 2304.0 | 3.0.0 |
78
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | COCO-Person | Int8 | 224x224x3 | STM32N6 | 1239.04 | 0.0 | 2304.0 | 3.0.0 |
79
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | COCO-Person | Int8 | 256x256x3 | STM32N6 | 1607.68 | 0.0 | 2304.0 | 3.0.0 |
 
80
 
81
 
82
  ### Reference **NPU** inference time based on COCO Person dataset (see Accuracy for details on dataset)
83
+ | Model | Dataset | Format | Resolution | Board | Execution Engine | Inference time (ms) | Inf / sec | STEdgeAI Core version |
84
+ |--------|------------------|--------|-------------|------------------|------------------|---------------------|-------|-------------------------|
85
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | COCO-Person | Int8 | 192x192x3 | STM32N6570-DK | NPU/MCU | 22.05 | 45.35 | 3.0.0 |
86
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | COCO-Person | Int8 | 224x224x3 | STM32N6570-DK | NPU/MCU | 27.64 | 36.18 | 3.0.0 |
87
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | COCO-Person | Int8 | 256x256x3 | STM32N6570-DK | NPU/MCU | 35.50 | 28.17 | 3.0.0 |
 
88
 
89
 
90
  ### Reference **MPU** inference time based on COCO Person dataset (see Accuracy for details on dataset)
91
+
92
+ | Model | Dataset | Format | Resolution | Quantization | Board | Execution Engine | Frequency | Inference time (ms) | %NPU | %GPU | %CPU | X-LINUX-AI version | Framework |
93
+ |--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|-----------------------------|--------|------------|----------------|-------------------|------------------|-----------|---------------------|-------|-------|------|--------------------|-----------------------|
94
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | custom_coco_person_17kpts | Int8 | 192x192x3 | per-channel** | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 55.81 | 2.87 | 97.13 | 0 | v6.1.0 | OpenVX |
95
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | custom_coco_person_17kpts | Int8 | 224x224x3 | per-channel** | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 79.41 | 2.41 | 97.59 | 0 | v6.1.0 | OpenVX |
96
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | custom_coco_person_17kpts | Int8 | 256x256x3 | per-channel** | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 68.42 | 3.32 | 96.68 | 0 | v6.1.0 | OpenVX |
97
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 192x192x3 | per-tensor | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 8.20 | 82.06 | 17.94 | 0 | v6.1.0 | OpenVX |
98
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 224x224x3 | per-tensor | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 11.63 | 83.75 | 16.25 | 0 | v6.1.0 | OpenVX |
99
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 256x256x3 | per-tensor | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 13.10 | 81.39 | 18.61 | 0 | v6.1.0 | OpenVX |
100
+ | [MoveNet Lightning](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_a100_192/movenet_singlepose_lightning_a100_192_int8.tflite) | custom_dataset_person_17kpts| Int8 | 192x192x3 | per-channel** | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 63.80 | 6.58 | 93.42 | 0 | v6.1.0 | OpenVX |
101
+ | [MoveNet Thunder](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_thunder_a100_256/movenet_singlepose_thunder_a100_256_int8.tflite) | custom_dataset_person_17kpts| Int8 | 256x256x3 | per-channel** | STM32MP257F-EV1 | NPU/GPU | 800 MHz | 183.49 | 3.47 | 96.53 | 0 | v6.1.0 | OpenVX |
102
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | custom_coco_person_17kpts | Int8 | 192x192x3 | per-channel | STM32MP157F-DK2 | 2 CPU | 800 MHz | 315.44 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
103
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | custom_coco_person_17kpts | Int8 | 224x224x3 | per-channel | STM32MP157F-DK2 | 2 CPU | 800 MHz | 416.98 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
104
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | custom_coco_person_17kpts | Int8 | 256x256x3 | per-channel | STM32MP157F-DK2 | 2 CPU | 800 MHz | 533.61 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
105
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 192x192x3 | per-tensor | STM32MP157F-DK2 | 2 CPU | 800 MHz | 424.77 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
106
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 224x224x3 | per-tensor | STM32MP157F-DK2 | 2 CPU | 800 MHz | 558.26 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
107
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 256x256x3 | per-tensor | STM32MP157F-DK2 | 2 CPU | 800 MHz | 727.03 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
108
+ | [MoveNet Lightning](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_a100_192/movenet_singlepose_lightning_a100_192_int8.tflite) | custom_dataset_person_17kpts| Int8 | 192x192x3 | per-channel | STM32MP157F-DK2 | 2 CPU | 800 MHz | 196.81 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
109
+ | [MoveNet Thunder](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_thunder_a100_256/movenet_singlepose_thunder_a100_256_int8.tflite) | custom_dataset_person_17kpts| Int8 | 256x256x3 | per-channel | STM32MP157F-DK2 | 2 CPU | 800 MHz | 766.38 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
110
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | custom_coco_person_17kpts | Int8 | 192x192x3 | per-channel | STM32MP135F-DK | 1 CPU | 1000 MHz | 484.64 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
111
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | custom_coco_person_17kpts | Int8 | 224x224x3 | per-channel | STM32MP135F-DK | 1 CPU | 1000 MHz | 651.62 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
112
+ | [ST MoveNet Lightning heatmaps](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | custom_coco_person_17kpts | Int8 | 256x256x3 | per-channel | STM32MP135F-DK | 1 CPU | 1000 MHz | 844.89 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
113
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 192x192x3 | per-tensor | STM32MP135F-DK | 1 CPU | 1000 MHz | 578.72 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
114
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 224x224x3 | per-tensor | STM32MP135F-DK | 1 CPU | 1000 MHz | 772.76 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
115
+ | [ST MoveNet Lightning heatmaps per-tensor](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8_per_tensor.tflite) | custom_coco_person_17kpts | Int8 | 256x256x3 | per-tensor | STM32MP135F-DK | 1 CPU | 1000 MHz | 1007.57 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
116
+ | [MoveNet Lightning](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_a100_192/movenet_singlepose_lightning_a100_192_int8.tflite) | custom_dataset_person_17kpts| Int8 | 192x192x3 | per-channel | STM32MP135F-DK | 1 CPU | 1000 MHz | 306.34 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
117
+ | [MoveNet Thunder](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_thunder_a100_256/movenet_singlepose_thunder_a100_256_int8.tflite) | custom_dataset_person_17kpts| Int8 | 256x256x3 | per-channel | STM32MP135F-DK | 1 CPU | 1000 MHz | 1131.30 | NA | NA | 100 | v6.1.0 | TensorFlowLite 2.18.0 |
118
 
119
  ** **To get the most out of MP25 NPU hardware acceleration, please use per-tensor quantization**
120
 
121
+ ** **Note:** On STM32MP2 devices, per-channel quantized models are internally converted to per-tensor quantization by the compiler using an entropy-based method. This may introduce a slight loss in accuracy compared to the original per-channel models.
122
+
123
  ### OKS on COCO Person dataset
124
 
125
 
126
  Dataset details: [link](https://cocodataset.org/#download) , License [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/legalcode) , Quotation[[1]](#1) , Number of classes: 80, Number of images: 118,287
127
 
128
 
129
+ | Model | Format | Resolution | Training Dataset | OKS |
130
+ |-------|--------|------------|------------------|----------------|
131
+ | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | Int8 | 192x192x3 | custom ST | 57.64 % |
132
+ | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | Int8 | 224x224x3 | custom ST | 62.29 % |
133
+ | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | Int8 | 256x256x3 | custom ST | 62.36 % |
134
+ | [ST MoveNet Lightning heatmaps per-tensor ](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8_per_tensor.tflite) | Int8 | 192x192x3 | custom ST | 55.84 % |
135
+ | [ST MoveNet Lightning heatmaps per-tensor ](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8_per_tensor.tflite) | Int8 | 224x224x3 | custom ST | 58.95 % |
136
+ | [ST MoveNet Lightning heatmaps per-tensor ](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_custom_dataset/custom_coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8_per_tensor.tflite) | Int8 | 256x256x3 | custom ST | 60.73 % |
137
+ | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8.tflite) | Int8 | 192x192x3 | COCO | 55.34 % |
138
+ | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8.tflite) | Int8 | 224x224x3 | COCO | 59.02 % |
139
+ | [ST MoveNet Lightning heatmaps per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8.tflite) | Int8 | 256x256x3 | COCO | 61.99 % |
140
+ | [ST MoveNet Lightning heatmaps per-tensor ](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/coco_person_17kpts/st_movenet_lightning_a100_heatmaps_192/st_movenet_lightning_a100_heatmaps_192_int8_per_tensor.tflite) | Int8 | 192x192x3 | COCO | 55.34 % |
141
+ | [ST MoveNet Lightning heatmaps per-tensor ](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/coco_person_17kpts/st_movenet_lightning_a100_heatmaps_224/st_movenet_lightning_a100_heatmaps_224_int8_per_tensor.tflite) | Int8 | 224x224x3 | COCO | 58.50 % |
142
+ | [ST MoveNet Lightning heatmaps per-tensor ](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/ST_pretrainedmodel_public_dataset/coco_person_17kpts/st_movenet_lightning_a100_heatmaps_256/st_movenet_lightning_a100_heatmaps_256_int8_per_tensor.tflite) | Int8 | 256x256x3 | COCO | 61.63 % |
143
+ | [MoveNet Lightning per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_lightning_a100_192/movenet_singlepose_lightning_a100_192_int8.tflite) | Int8 | 192x192x3 | custom Google | 54.12 % |
144
+ | [MoveNet Thunder per-channel](https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/pose_estimation/movenet/Public_pretrainedmodel_custom_dataset/custom_dataset_person_17kpts/movenet_thunder_a100_256/movenet_singlepose_thunder_a100_256_int8.tflite) | Int8 | 256x256x3 | custom Google | 64.43 % |
145
 
 
146
 
147
  ## Integration in a simple example and other services support:
148