Spaces:
Runtime error
Runtime error
Upload object_detection.py
Browse files- object_detection.py +27 -0
object_detection.py
ADDED
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# from PIL import Image
|
2 |
+
from transformers import DetrFeatureExtractor
|
3 |
+
from transformers import DetrForObjectDetection
|
4 |
+
import torch
|
5 |
+
# import numpy as np
|
6 |
+
|
7 |
+
def object_count(picture):
|
8 |
+
|
9 |
+
feature_extractor = DetrFeatureExtractor.from_pretrained("facebook/detr-resnet-50")
|
10 |
+
encoding = feature_extractor(picture, return_tensors="pt")
|
11 |
+
model = DetrForObjectDetection.from_pretrained("facebook/detr-resnet-50")
|
12 |
+
outputs = model(**encoding)
|
13 |
+
# keep only predictions of queries with 0.9+ confidence (excluding no-object class)
|
14 |
+
probas = outputs.logits.softmax(-1)[0, :, :-1]
|
15 |
+
keep = probas.max(-1).values > 0.7
|
16 |
+
count = 0
|
17 |
+
for i in keep:
|
18 |
+
if i:
|
19 |
+
count=count+1
|
20 |
+
|
21 |
+
return "About " + str(count) +" common objects were detected"
|
22 |
+
|
23 |
+
# object_count("toothbrush.jpg")
|
24 |
+
import gradio as gr
|
25 |
+
|
26 |
+
interface = gr.Interface(object_count, gr.inputs.Image(shape=(640, 480)), "text").launch()
|
27 |
+
|