Open ilie321 opened 1 year ago
Sorry I can't help you with that. In fact the work I have done is very small. As the readme shows, these projects have original projects (use Python language and .pth models), nihui produced inference in c language and ncnn models. This repository uses nihui's results directly. Currently there are tools to directly convert pth models to onnx models, but the inference need to be rewritten. I have no such experience.
Ok thank you. Do you know what settings are better for ncnn for video conversion? Like the net.opt are different?
Anyone? Thanks
A model processing video is actually to parse the video into multiple pictures, then process each image separately, and finally connect these pictures into a video again. For ir sensors, you may need to train a dedicated model. There are many problems with existing models, such as the image resolution of the model is higher than that of ir sensor, but I am not good at these so could not help you
Thanks for answering. Using your models i got good results for IR but i need for pcb to get good results and this is my problem. Can you point me where to start for build a good model step by step tha is working with ncnn? I found some pytorch gits and i tried to train some model with ir images but image is distorted and only one color and after converting pth to ncnn is give me black screen.
Yor models example : 80x60 image
Results look like this :amazing
I'm not sure which model you used? realesrgan pth model to ncnn model: https://github.com/tumuyan/RealSR-NCNN-Android#add-more-models-to-realsr-ncnn-android-gui
Hello i have a free app with code for thermal camera and i want to try the models you have in this git but the problems is that my app use opencv which know only .pb or .onnx models and i'm stuck can you help me somehow? thank you