Recognizing the wrong person

[rafaelsilvade]

Issue Description Hi,

After several tests and I managed to capture the evidence.

As you can see, people are different and yet the model puts a similarity of 0.74. I tested on other sites and models and even the facejs version below your lib doesn't match the two photos.

I am attaching the two photos, for reasons of privacy for both people, could I not leave the images public?

Steps to Reproduce

compare the similarity between the two images

Expected Behavior

Environment humanConfig = { debug: false, modelBasePath: '../assets/models/', filter: { enabled: true, equalization: true, flip: false, width: 0 }, face: { enabled: true, detector: { rotation: true, maxDetected: 1, minConfidence: 0.6, return: true, mask: false, skipFrames: 1, iouThreshold: 0.5 }, iris: { enabled: true }, description: { enabled: true, minConfidence: 0.6, skipFrames: 1 }, // default model for face descriptor extraction is faceres

  //insightface: { enabled: true, modelPath: '../assets/models/insightface-mobilenet-swish.json' }, // alternative model
  mesh: { enabled: true },
  emotion: { enabled: false },
  antispoof: { enabled: false },
  liveness: { enabled: false },
  skipFrames: 1
},
body: { enabled: false },
hand: { enabled: false },
object: { enabled: false },
gesture: { enabled: true },
segmentation: { enabled: false },
async: false

};

• Human library version? 3.2.1
• Built-in demo or custom code?
• Type of module used (e.g. js, esm, esm-nobundle)?
• TensorFlow/JS version (if not using bundled module)?
• Browser or NodeJS and version (e.g. NodeJS 14.15 or Chrome 89)? Chrome
• OS and Hardware platform (e.g. Windows 10, Ubuntu Linux on x64, Android 10)? Windows 11
• Packager (if any) (e.g, webpack, rollup, parcel, esbuild, etc.)?
• Framework (if any) (e.g. React, NextJS, etc.)? Angular

Diagnostics

• Check out any applicable diagnostic steps

Additional

• For installation or startup issues include your package.json
• For usage issues, it is recommended to post your code as gist
• For general questions, create a discussion topic

[vladmandic]

this topic has been discussed many times:

• search for more detailed descriptions/conversations - there have been a lot.
• default model for face intentification is choosen because its fast and lightweight,
  not because its precise. use a different model, human supports several just via config change
• tune other params as needed

and other than those 3 large topics, i really don't have much to add so i'm closing the issue for now.