Face Detection and Facial Features

FREEMIUM
Verified
Por Imagevision | Atualizado vor einem Jahr | Machine Learning
Popularidade

8.8 / 10

Latência

2,723ms

Nível de serviço

98%

Voltar para todas as discussões

Endpoint for detecting throws Internal Server Error

avatar
ljubisap
vor 2 Jahren

Hi,

I wanted to try out the API and I got Internal Server Error.
The API returns the same for the provided image URL and I tried out for some other image.

Can you please check what is the problem?

Best,
Ljubisa.

avatar
necococo commented vor 2 Jahren

How can I use these values of json to calculate the similarity of two faces?

Also, are the featureDistances normalized in size?

I used a picture of a cat.

2.2 was the largest.

How did you calculate the distance?

avatar
ljubisap commented vor 2 Jahren

Thank you @eyerecognize.
I assumed it works that way.

avatar
eyerecognize commented vor 2 Jahren

The values returned represent ratios with respect to your original image with a value between 0 and 1. You can multiply this number by your image size to get the bounding box coordinates. For Example, if your original image was 640 x 480:

Left coordinate = BoundingBox.Left (0.3922065) * image width (640) = 251

Top coordinate = BoundingBox.Top (0.15567766) * image height (480) = 75

Face width = BoundingBox.Width (0.284666) * image width (640) = 182

Face height = BoundingBox.Height (0.2930403) * image height (480) = 140

You can then use the above to draw bounding box like this (Python PIL library)

draw.rectangle([left,top, left + width, top + height], outline=’#00d400’)

avatar
ljubisap commented vor 2 Jahren

Thank you @eyerecognize for your quick response.

One question, I got a response like this:
{ "version":"1.002.03", "objecturl":"https://media.kairos.com/kairos-elizabeth.jpg", "message":"success", "resource":"Deep Face Detect in Images by EyeRecognize", "FaceDetails":[ 0:{ "BoundingBox":{ "Width":0.4896048605442047, "Height":0.5420181751251221, "Left":0.2524735927581787, "Top":0.19791695475578308 }, "Landmarks":[ 0:{ "Type":"eyeLeft", "X":0.3848588466644287, "Y":0.4365363121032715 }, 1:{ "Type":"eyeRight", "X":0.6189132332801819, "Y":0.4381250739097595 }, 2:{ "Type":"mouthLeft", "X":0.4019031524658203, "Y":0.6177037954330444 }, 3:{ "Type":"mouthRight", "X":0.597312867641449, "Y":0.6189464926719666 }, 4:{ "Type":"nose", "X":0.5029631853103638, "Y":0.5462852120399475 } ], "Pose":{ "Roll":-0.031106360256671906, "Yaw":0.836026132106781, "Pitch":-2.362826108932495 }, "Quality":{ "Brightness":67.24449157714844, "Sharpness":83.14741516113281 }, "Confidence":99.99556732177734 } ] }

For all X/Y positions, I got numbers less than 1. Does it mean that the location is relative according to the image size (from 0 to 1)?

Also for nose location, I get a single point in response, are there additional attributes, like nose width/height etc.?

Best.

avatar
eyerecognize commented vor 2 Jahren
avatar
ljubisap commented vor 2 Jahren

No idea why images won’t show up here. Here is the URL https://imgur.com/a/x2W45oN

Junte-se à discussão - adicione o comentário abaixo:

Efetue login / inscreva-se para postar novos comentários
Avaliação: 1 - Votos: 1