Node.js调用GC Vision annotateImage遇两类错误：Feature未定义及类型不存在

Hey there! Let's work through those errors you're hitting when trying to pull more than 10 web detection results with Google Cloud Vision in Node.js.

Let's break down the root causes first: #

1. Cannot read property 'Feature' of undefined: This error pops up because the way you're referencing the feature type (vision.types.Feature.Type.WEB_DETECTION) doesn't match the current version of the @google-cloud/vision client library. Newer releases changed how these enums are exported, so that path no longer exists.
2. annotateImage feature types don't exist: This is a secondary issue caused by the invalid type reference—since the library can't recognize the feature type you're passing, it throws this validation error.

Here's your fixed code (most compatible approach): #

Instead of relying on the broken enum path, use the direct string value for the feature type. This works reliably across all recent library versions:

const vision = require('@google-cloud/vision');
const client = new vision.ImageAnnotatorClient();

const webSearchRequest = {
  image: { source: { imageUri: `gs://${bucket.name}/${filePath}` } },
  features: [{ maxResults: 50, type: 'WEB_DETECTION' }] // Use string type directly
};

return client.annotateImage(webSearchRequest)
  .then(webResults => {
    console.log('Web detection results:', webResults);
  })
  .catch(err => {
    console.error('Error fetching results:', err);
  });

If you prefer using enums (optional): #

If you want to stick with enum values instead of strings, you can reference the proto definitions directly (this works for v2+ of the library):

const vision = require('@google-cloud/vision');
const client = new vision.ImageAnnotatorClient();
// Import the Feature enum from the official proto definitions
const { Feature } = vision.protos.google.cloud.vision.v1;

const webSearchRequest = {
  image: { source: { imageUri: `gs://${bucket.name}/${filePath}` } },
  features: [{ maxResults: 50, type: Feature.Type.WEB_DETECTION }]
};

return client.annotateImage(webSearchRequest)
  .then(webResults => {
    console.log('Web detection results:', webResults);
  })
  .catch(err => {
    console.error('Error fetching results:', err);
  });

Quick tips: #

• Always add a .catch() block to handle unexpected errors—your original code was missing this, which would make debugging way harder.
• Using the string type ('WEB_DETECTION') is more future-proof, as enum export paths can shift between library updates.

内容的提问来源于stack exchange，提问作者cerealex