Update app.py

app.py

@@ -1,50 +1,57 @@
-import mxnet as mx
-import matplotlib.pyplot as plt
+import onnx
 import numpy as np
-from collections import namedtuple
-from mxnet.gluon.data.vision import transforms
-import os
-import gradio as gr
-
-from PIL import Image
-import imageio
 import onnxruntime as ort
-from torchvision import transforms
-
+from PIL import Image
+import cv2
 
-preprocess = transforms.Compose([
-    transforms.Resize(256),
-    transforms.CenterCrop(224),
-    transforms.ToTensor(),
-    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
-])
+os.system("wget https://s3.amazonaws.com/onnx-model-zoo/synset.txt")
 
-mx.test_utils.download('https://s3.amazonaws.com/model-server/inputs/kitten.jpg')
 
-mx.test_utils.download('https://s3.amazonaws.com/onnx-model-zoo/synset.txt')
 with open('synset.txt', 'r') as f:
     labels = [l.rstrip() for l in f]
     
 os.system("wget https://github.com/AK391/models/raw/main/vision/classification/densenet-121/model/densenet-9.onnx")
 
-ort_session = ort.InferenceSession("densenet-9.onnx")
+os.system("wget https://s3.amazonaws.com/model-server/inputs/kitten.jpg")
 
 
-def predict(pil):
-    input_tensor = preprocess(pil)
-    img_batch = input_tensor.unsqueeze(0) # create a mini-batch as expected by the model
-    img_batch_np = img_batch.cpu().detach().numpy()
 
-    outputs = ort_session.run(
-        None,
-        {"data_0": img_batch_np.astype(np.float32)},
-    )
+model_path = 'resnet50-v1-12.onnx'
+model = onnx.load(model_path)
+session = ort.InferenceSession(model.SerializeToString())
 
-    a = np.argsort(-outputs[0].flatten())
+def get_image(path, show=False):
+    with Image.open(path) as img:
+        img = np.array(img.convert('RGB'))
+    if show:
+        plt.imshow(img)
+        plt.axis('off')
+    return img
+    
+def preprocess(img):
+    img = img / 255.
+    img = cv2.resize(img, (256, 256))
+    h, w = img.shape[0], img.shape[1]
+    y0 = (h - 224) // 2
+    x0 = (w - 224) // 2
+    img = img[y0 : y0+224, x0 : x0+224, :]
+    img = (img - [0.485, 0.456, 0.406]) / [0.229, 0.224, 0.225]
+    img = np.transpose(img, axes=[2, 0, 1])
+    img = img.astype(np.float32)
+    img = np.expand_dims(img, axis=0)
+    return img
+
+def predict(path):
+    img = get_image(path, show=True)
+    img = preprocess(img)
+    ort_inputs = {session.get_inputs()[0].name: img}
+    preds = session.run(None, ort_inputs)[0]
+    preds = np.squeeze(preds)
+    a = np.argsort(preds)[::-1]
     results = {}
-    for i in a[0:5]:
-        results[labels[i]]=float(outputs[0][0][i])
+    results[labels[a[0]]] = preds[a[0]]
     return results
+
        
 
 title="DenseNet-121"