Coincidentally, all these problems are solved in hcaptcha-challenger.🤷♀️
The recognition task is done independently and not with the help of extensions, so the browser can be started headless.
The recognition module uses a self-trained yolov5n6(onnx) to perform two challenging rounds of the recognition task in one second in extreme cases.This model has excellent penetration of the hcaptcha challenge, is less than 15MB in size, and can run on almost any embedded device.
In the following demo, I am using yolov5s6(onnx) with a larger parameter size to start the solution. if you specify yolov5n6(onnx) to start, the detection speed can be twice as fast...🤦♂️
To answer the title Can I use the hcaptcha test site you built, yes you can :)
This repository is no longer maintained and, I know, is very slow compared to image recognition (Yolo v5s6).
Hi, I used the
hcaptcha
test site you built in the hcaptcha-challenger project I created recently.In addition, I noticed that the performance issues were mentioned in the previous discussion of this extension.
Coincidentally, all these problems are solved in
hcaptcha-challenger
.🤷♀️In the following demo, I am using yolov5s6(onnx) with a larger parameter size to start the solution. if you specify yolov5n6(onnx) to start, the detection speed can be twice as fast...🤦♂️