Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -9,7 +9,8 @@ import matplotlib.pyplot as plt
|
|
9 |
import numpy as np
|
10 |
from sklearn.preprocessing import LabelEncoder
|
11 |
from huggingface_hub import hf_hub_download
|
12 |
-
from transformers import
|
|
|
13 |
from sklearn.feature_extraction.text import TfidfVectorizer
|
14 |
from sklearn.metrics.pairwise import cosine_similarity
|
15 |
import re
|
@@ -31,14 +32,28 @@ def load_image(image_file):
|
|
31 |
|
32 |
def classify_image(image):
|
33 |
try:
|
34 |
-
#
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
|
40 |
-
#
|
41 |
-
|
|
|
|
|
|
|
|
|
42 |
|
43 |
except Exception as e:
|
44 |
st.error(f"Classification error: {e}")
|
|
|
9 |
import numpy as np
|
10 |
from sklearn.preprocessing import LabelEncoder
|
11 |
from huggingface_hub import hf_hub_download
|
12 |
+
from transformers import AutoFeatureExtractor, AutoModelForImageClassification
|
13 |
+
import torch
|
14 |
from sklearn.feature_extraction.text import TfidfVectorizer
|
15 |
from sklearn.metrics.pairwise import cosine_similarity
|
16 |
import re
|
|
|
32 |
|
33 |
def classify_image(image):
|
34 |
try:
|
35 |
+
# Load the model and feature extractor
|
36 |
+
model_name = "dima806/car_models_image_detection"
|
37 |
+
feature_extractor = AutoFeatureExtractor.from_pretrained(model_name)
|
38 |
+
model = AutoModelForImageClassification.from_pretrained(model_name)
|
39 |
+
|
40 |
+
# Preprocess the image
|
41 |
+
inputs = feature_extractor(images=image, return_tensors="pt")
|
42 |
+
|
43 |
+
# Perform inference
|
44 |
+
with torch.no_grad():
|
45 |
+
outputs = model(**inputs)
|
46 |
+
|
47 |
+
# Get the predicted class
|
48 |
+
logits = outputs.logits
|
49 |
+
predicted_class_idx = logits.argmax(-1).item()
|
50 |
|
51 |
+
# Get the class label and score
|
52 |
+
predicted_class_label = model.config.id2label[predicted_class_idx]
|
53 |
+
score = torch.nn.functional.softmax(logits, dim=-1)[0, predicted_class_idx].item()
|
54 |
+
|
55 |
+
# Return the top prediction
|
56 |
+
return [{'label': predicted_class_label, 'score': score}]
|
57 |
|
58 |
except Exception as e:
|
59 |
st.error(f"Classification error: {e}")
|