Error executing remote API: Extra data: line 1 column 329 (char 328)

[mfauvain]

Hi, not sure where to start here. I don't know when those errors have started to show up, as it fallback to server when the mlapi server is down, I didn't notice up until now.. I've reinstalled mlapi but no luck.

[mfauvain]

Adding debug log debug.txt

zmevent doesn't seem to 'connect' to the mlapi server but I can't understand why..

[pliablepixels]

I'm not sure either. Works for me. Try to update your pyzm library to the latest version. Also see what logs mlapi is printing

[mfauvain]

I've reinstalled pyzm on the mlapi server and zmeventserver, no luck. However I tried to shutdown the mlapi server and the zmevent server gives the same error so it does not seem to be on the mlapi server side. I've reinstalled zmeventservernotication but same error on the zmevent server side. on mlapi server no error reported, has it seems no connection is even made.. It really look as if the zmeventserver does not 'connect' for some reason.

EDIT : for avoidance of doubt, I've run a sample object detect on the mlapi server (as specified in the install examples) and it correctly detect a bird. So clearly coming from the zmeventserver.

[mfauvain]

ah. found an old bug report (of mine actually) on the zmeventnotification git page where you told me to delete zm_login.json on the server and ... it worked again.

[gfro84]

I hate holding back on asking a question, then no sooner do you ask it, you figure out the answer yourself!

How do you find the coral library compared to yolo? I liked the speed but yolo on cpu gives me more accurate results, and I didn't really have the need for speed so sent it back.

On Sun, 13 Dec 2020, 21:10 Marc, notifications@github.com wrote:

ah. found an old bug report (of mine actually) on the zmeventnotification git page where you told me to delete zm_login.json on the server and ... it worked again.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/pliablepixels/mlapi/issues/24#issuecomment-744068940, or unsubscribe https://github.com/notifications/unsubscribe-auth/AG6VOJG3LZF5O6IGDMOSMCLSUUUTZANCNFSM4UYODWEQ .

[pliablepixels]

Faster but less accurate. However check out the work @rabsym has done on retraining models. He did It for yolo but the same approach applies to coral models too.

[mfauvain]

@gfro84 : indeed... did not try CORAL yet. I am happy with yolo though.

[pliablepixels]

closing as the core issue seems to be resolved.