| --- |
| library_name: tflite |
| pipeline_tag: image-classification |
| tags: |
| - image-classification |
| - tflite |
| - tensorflow |
| - gender-classification |
| base_model: rizvandwiki/gender-classification-2 |
| --- |
| |
| # Gender classification (TensorFlow / TFLite) |
|
|
| TensorFlow and TFLite conversion of |
| [rizvandwiki/gender-classification-2](https://huggingface.co/rizvandwiki/gender-classification-2) |
| (ViT). Please check the original model card for license and training details. |
|
|
| ## Files |
| - `gender_model.tflite` : mobile model (float16) |
| - `tf_model.h5` : TensorFlow (Keras) weights |
| - `labels.txt` : 0 = female, 1 = male |
|
|
| ## Input / preprocessing |
| - Shape: `[1, 3, 224, 224]` (channels first, RGB) |
| - Normalize: `(pixel / 255 - 0.5) / 0.5` |
| - Detect and crop the face before running the model. |
| - Output: 2 logits -> apply softmax. |
|
|
| Note: converted with SELECT_TF_OPS, so on Android add |
| `org.tensorflow:tensorflow-lite-select-tf-ops`. |
|
|
| ## Python example |
| ```python |
| import numpy as np, tensorflow as tf |
| from huggingface_hub import hf_hub_download |
| |
| path = hf_hub_download("SynthAIzer/gender-classification-tflite", "gender_model.tflite") |
| interp = tf.lite.Interpreter(model_path=path) |
| interp.allocate_tensors() |
| ``` |
|
|
