| language: en | |
| license: apache-2.0 | |
| tags: | |
| - axera | |
| - ax650 | |
| - axmodel | |
| - image-classification | |
| - mobilenet | |
| - timm | |
| library_name: axengine | |
| pipeline_tag: image-classification | |
| # MobileNetV3-Small — AX650 Image Classification | |
| MobileNetV3-Small (ImageNet-1k, 1000 classes) compiled to AX650 AXMODEL via Pulsar2. | |
| ## Model | |
| | Item | Value | | |
| |------|-------| | |
| | Architecture | MobileNetV3-Small 100 | | |
| | Source | [timm/mobilenetv3_small_100.lamb_in1k](https://hf-mirror.com/timm/mobilenetv3_small_100.lamb_in1k) | | |
| | Task | Image Classification (1000 cls) | | |
| | Input | 224×224 BGR, uint8→float [0,1] | | |
| | Chip | AX650N (NPU3) | | |
| | Quantization | INT8 | | |
| | Size | 3.3 MB | | |
| | Board | BSP 3.10.2, axengine.InferenceSession | | |
| ## Usage (on AX650 board) | |
| ```python | |
| import numpy as np | |
| import axengine | |
| sess = axengine.InferenceSession("model.axmodel") | |
| data = np.random.rand(1, 3, 224, 224).astype(np.float32) | |
| out = sess.run(None, {"images": data}) | |
| print(out[0].argmax()) # predicted class | |
| ``` | |