Medric Sonwa

Hi ! Thanks a lot for this repo ! I have encountered the same problem stated at the beginning of this issue. When I try to get the torch script...

It seems that for some methods like `camera.take_photo()` `camera.set_mode()` or `camera.format_storage()`, the communication goes well. But for others (particularly for those that return an answer) like `camera.status()`, `camera.capture_info()` or `camera.list_photos()`,...

I see. No, actually QGC streams perfectly the drone camera.

Did you succeed to get photos with another drone model? Which one please?

Okay I see. Finally, since my initial objective was simply to take pictures, regardless of the methods, I switched to this method ``` cam = cv2.VideoCapture("udpsrc port=5600 ! application/x-rtp, payload=96...

Finally, I fixed the problem by setting the value "bytes" to the parameter **encoding** in the function pickle.load(), like that : `qnn = pickle.load(open("mnist-w1a1.pickle", "rb"), encoding="bytes")`

> > Finally, I fixed the problem by setting the value "bytes" to the parameter **encoding** in the function pickle.load(), like that : > > `qnn = pickle.load(open("mnist-w1a1.pickle", "rb"), encoding="bytes")`...

The problem is that, when the data of the neural network are loaded, this one doesn't consider the attributes created in its constructor, in fact these elements have not been...

@Nuno-Mota I have the same issue, but this time with FasterRCNN. Have you found a solution, please?