blazepalm
blazepalm copied to clipboard
Window resize + Detection problem
I managed to make it work somehow by fixing my Tensorflow GPU issues.
However, the current solution only works for a window having 256 x 256, and changing the dimensions on codebase doesn't work. Also, the detections seem inaccurate like the snap below:
If improvements are needed on this, I'll wait, or I'll explore. Thanks!
Yes I believe many improvements are required for this repo (including a dynamic input size). My attention and time has been pulled to other things, so I have not been working on it. I may create a similar repo for pytorch rather than TF if one does not already exist. Although do not think I will continue with this TF implementation.
limão calamitoso, Poderia explicar como executou o algoritmo ?