Update validated TorchVision exports and compatibility documentation

Publish the reviewed model files and current CPU/GPU/NPU compatibility table.

README.md

@@ -35,6 +35,15 @@ model-index:
 
 EfficientNet B0 model pre-trained on ImageNet-1k. Originally introduced by Tan and Le in the influential paper, [**EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks**](https://arxiv.org/abs/1905.11946) this model utilizes compound scaling to systematically balance network depth, width, and resolution, enabling superior accuracy with significantly higher efficiency than traditional architectures.
 
+`efficientnet_b0_int8_channelwise.tflite`: Mixed INT8/FP32 with channelwise INT8 weights. The prefix through stage 5 remains FP32.
+
+## Compatibility
+
+| File | CPU | GPU | NPU |
+|---|---|---|---|
+| `efficientnet_b0.tflite` | Supported | Supported | N/A |
+| `efficientnet_b0_int8_channelwise.tflite` | Supported | Not supported | Qualcomm / MediaTek |
+
 ## Intended uses & limitations
 
 The model files were converted from pretrained weights from PyTorch Vision. The models may have their own licenses or terms and conditions derived from PyTorch Vision and the dataset used for training. It is your responsibility to determine whether you have permission to use the models for your use case.
@@ -83,18 +92,18 @@ def preprocess(img: Image.Image) -> np.ndarray:
     w, h = img.size
     s = 256
     if w < h:
-        img = img.resize((s, int(round(h * s / w))), Image.BICUBIC)
+        img = img.resize((s, int(h * s / w)), Image.BICUBIC)
     else:
-        img = img.resize((int(round(w * s / h)), s), Image.BICUBIC)
-    left = (img.size[0] - 224) // 2
-    top = (img.size[1] - 224) // 2
+        img = img.resize((int(w * s / h), s), Image.BICUBIC)
+    left = int(round((img.size[0] - 224) / 2.0))
+    top = int(round((img.size[1] - 224) / 2.0))
     img = img.crop((left, top, left + 224, top + 224))
 
     x = np.asarray(img, dtype=np.float32) / 255.0
     x = (x - np.array([0.485, 0.456, 0.406], dtype=np.float32)) / np.array(
         [0.229, 0.224, 0.225], dtype=np.float32
     )
-    return np.transpose(x, (2, 0, 1))
+    return np.ascontiguousarray(x.transpose(2, 0, 1)[None])
 
 def main():
     ap = argparse.ArgumentParser()

efficientnet_b0.tflite

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aefc589950121af9bc3288f9118337719f8e0f1dc70ebea179609521127c5d0d
-size 21319488
+oid sha256:9b58d40a42fc7996070de6256a3a772771885a2609fc97753d054a3e6ea1ea98
+size 21126448

efficientnet_b0_int8_channelwise.tflite

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5106e5665786e750cae6c90bc0710410bb0ec34b20ed0d8cff5c54b334498c98
+size 8383640