Instructions to use Chizzsam/shape-models-checkpoints with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Keras
How to use Chizzsam/shape-models-checkpoints with Keras:
# !pip install -U keras tensorflow huggingface_hub # Keras needs TensorFlow installed to read "hf://" paths, so the tensorflow backend is selected here; # "jax" and "torch" also work for computation once TensorFlow is installed. import os os.environ["KERAS_BACKEND"] = "tensorflow" import keras model = keras.saving.load_model("hf://Chizzsam/shape-models-checkpoints") - Notebooks
- Google Colab
- Kaggle
Upload body/body_shape_norm.json with huggingface_hub
Browse files- body/body_shape_norm.json +97 -0
body/body_shape_norm.json
ADDED
|
@@ -0,0 +1,97 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_version": "3.2.0",
|
| 3 |
+
"feature_names": [
|
| 4 |
+
"shoulder_hip_ratio",
|
| 5 |
+
"waist_hip_ratio",
|
| 6 |
+
"waist_shoulder_ratio",
|
| 7 |
+
"bust_hip_ratio",
|
| 8 |
+
"bust_waist_ratio",
|
| 9 |
+
"shoulder_waist_diff",
|
| 10 |
+
"hip_shoulder_diff",
|
| 11 |
+
"hip_waist_diff"
|
| 12 |
+
],
|
| 13 |
+
"mean": [
|
| 14 |
+
0.37780097126960754,
|
| 15 |
+
0.8161380887031555,
|
| 16 |
+
2.200714111328125,
|
| 17 |
+
0.9687443375587463,
|
| 18 |
+
1.1990960836410522,
|
| 19 |
+
-44.1546516418457,
|
| 20 |
+
62.62126541137695,
|
| 21 |
+
18.466611862182617
|
| 22 |
+
],
|
| 23 |
+
"std": [
|
| 24 |
+
0.05251329392194748,
|
| 25 |
+
0.0854557454586029,
|
| 26 |
+
0.3728339374065399,
|
| 27 |
+
0.09726245701313019,
|
| 28 |
+
0.17114204168319702,
|
| 29 |
+
11.966014862060547,
|
| 30 |
+
10.920693397521973,
|
| 31 |
+
9.034283638000488
|
| 32 |
+
],
|
| 33 |
+
"raw_min": [
|
| 34 |
+
0.2409999966621399,
|
| 35 |
+
0.611299991607666,
|
| 36 |
+
1.417099952697754,
|
| 37 |
+
0.6786999702453613,
|
| 38 |
+
0.8282999992370605,
|
| 39 |
+
-76.9000015258789,
|
| 40 |
+
38.099998474121094,
|
| 41 |
+
-4.400000095367432
|
| 42 |
+
],
|
| 43 |
+
"raw_max": [
|
| 44 |
+
0.5447999835014343,
|
| 45 |
+
1.045799970626831,
|
| 46 |
+
3.4370999336242676,
|
| 47 |
+
1.2482999563217163,
|
| 48 |
+
1.700600028038025,
|
| 49 |
+
-18.100000381469727,
|
| 50 |
+
95.0999984741211,
|
| 51 |
+
41.400001525878906
|
| 52 |
+
],
|
| 53 |
+
"normalized_min": [
|
| 54 |
+
-2.6050732135772705,
|
| 55 |
+
-2.397007942199707,
|
| 56 |
+
-2.101778030395508,
|
| 57 |
+
-2.982079267501831,
|
| 58 |
+
-2.1665985584259033,
|
| 59 |
+
-2.7365293502807617,
|
| 60 |
+
-2.245394706726074,
|
| 61 |
+
-2.531092882156372
|
| 62 |
+
],
|
| 63 |
+
"normalized_max": [
|
| 64 |
+
3.180128335952759,
|
| 65 |
+
2.687494993209839,
|
| 66 |
+
3.3161838054656982,
|
| 67 |
+
2.874239683151245,
|
| 68 |
+
2.930337429046631,
|
| 69 |
+
2.1773874759674072,
|
| 70 |
+
2.9740540981292725,
|
| 71 |
+
2.538484573364258
|
| 72 |
+
],
|
| 73 |
+
"class_names": [
|
| 74 |
+
"Hourglass",
|
| 75 |
+
"Pear",
|
| 76 |
+
"Apple",
|
| 77 |
+
"Rectangle",
|
| 78 |
+
"Inverted Triangle"
|
| 79 |
+
],
|
| 80 |
+
"input": {
|
| 81 |
+
"shape": [
|
| 82 |
+
1,
|
| 83 |
+
8
|
| 84 |
+
],
|
| 85 |
+
"dtype": "float32",
|
| 86 |
+
"note": "standardize raw ratios as (x - mean) / std BEFORE inference"
|
| 87 |
+
},
|
| 88 |
+
"output": {
|
| 89 |
+
"shape": [
|
| 90 |
+
1,
|
| 91 |
+
5
|
| 92 |
+
],
|
| 93 |
+
"dtype": "float32",
|
| 94 |
+
"type": "softmax"
|
| 95 |
+
},
|
| 96 |
+
"note_app_side": "The three *_diff features were trained on raw cm values. The app must divide each diff by hip width before standardizing so the feature becomes dimensionless and equivalent across px/cm scales."
|
| 97 |
+
}
|