我将Google Object Detection API用于我们的自定义模型。
数据集包含名片,因此所有文本。我们已经部署了模型,并且当我们使用可视界面测试名片时,模型运行得很好。
但是,要在后端Nodejs服务器上使用它,我们正在使用REST API。 API请求如下所示:
curl -X POST -H "Content-Type: application/json" -H "Authorization: Bearer $(gcloud auth application-default print-access-token)" https://automl.googleapis.com/v1beta1/projects/1023422831715/locations/us-central1/models/IOD9200669320764456960:predict -d @request.json
REST响应返回的数据如下:
{
"payload": [
{
"annotationSpecId": "5824016335306227712",
"imageObjectDetection": {
"boundingBox": {
"normalizedVertices": [
{
"x": 0.050029,
"y": 0.139873
},
{
"x": 0.230016,
"y": 0.251469
}
]
},
"score": 0.998103
},
"displayName": "first_name"
},
{
"annotationSpecId": "2224232858153648128",
"imageObjectDetection": {
"boundingBox": {
"normalizedVertices": [
{
"x": 0.0465549,
"y": 0.236178
},
{
"x": 0.462747,
"y": 0.30602
}
]
},
"score": 0.98034
},
"displayName": "job_title"
},
{
"annotationSpecId": "8910530192426926080",
"imageObjectDetection": {
"boundingBox": {
"normalizedVertices": [
{
"x": 0.053251,
"y": 0.410447
},
{
"x": 0.452525,
"y": 0.559461
}
]
},
"score": 0.904657
},
"displayName": "address"
}
]
}
问题在于,在REST情况下,这仅给出了带有分数和标签的边界框。但是,它没有给出。边框下的文字。
那么,我们如何在上述建议的边框下获得文本?