Views
No views yet
1import requests
2import base64
3
4API_URL = "https://your-endpoint.endpoints.huggingface.cloud"
5headers = {"Authorization": "Bearer YOUR_TOKEN"}
6
7# Load images as base64
8with open("screenshot.png", "rb") as f:
9 target_b64 = base64.b64encode(f.read()).decode()
10with open("icon.png", "rb") as f:
11 query_b64 = base64.b64encode(f.read()).decode()
12
13response = requests.post(API_URL, headers=headers, json={
14 "inputs": {
15 "target_image": target_b64,
16 "query_image": query_b64,
17 "threshold": 0.5
18 }
19})
20
21print(response.json())
22# {"detections": [{"box": [100, 200, 150, 250], "confidence": 0.92}]}1response = requests.post(API_URL, headers=headers, json={
2 "inputs": {
3 "target_image": target_b64,
4 "queries": ["a play button", "a settings icon"],
5 "threshold": 0.1
6 }
7})1response = requests.post(API_URL, headers=headers, json={
2 "inputs": {
3 "target_image": target_b64,
4 "query_images": [icon1_b64, icon2_b64, icon3_b64],
5 "threshold": 0.5
6 }
7})
8# Results include "label": "query_0", "query_1", etc.| Parameter | Type | Default | Description |
|---|---|---|---|
target_image | string | required | Base64-encoded target image |
query_image | string | - | Base64-encoded reference image |
query_images | array | - | Multiple base64-encoded reference images |
queries | array | - | Text descriptions to search for |
threshold | float | 0.5 | Confidence threshold (0-1) |
nms_threshold | float | 0.3 | Non-max suppression threshold |
1{
2 "detections": [
3 {
4 "box": [x1, y1, x2, y2],
5 "confidence": 0.95,
6 "label": "query_0"
7 }
8 ]
9}google/owlv2-large-patch14-ensemble for best accuracy.