Classifier.py and WebCam.py problems

[Mattispark]

Good morning Mr. Adam Milton-Barker
I am Prati Mattia, currently a trainee at Spark, an Italian company that deals with smart video cameras and security. I'm contacting you because, after following your guide (https://www.hackster.io/AdamMiltonBarker/facial-rec-with-movidius-siamese-nets-triplet-loss-d9cb4f) using a single ubuntu 16.04 machine, one of the program in the project (Classifier.py) returns me an output in my opinion incorrect. In particular, despite having inserted the dataset as you announced, the program simply analyzes and print only the elements of the testing folder returning a generic calculation of the inference (not communicating with which image of the known folder is comparing) .... all this, also due to the absence of a graphical interface, makes it impossible for me to understand the output and which comparisons have been successful or not
In addition, the program gives me a 100% match on every single image (eg: 31 visits and 31 successful matches) that I do not think is correct considering the quality of the images I have inserted ... not that I expect total absence or presence of match, but I hypothesized in a middle way

**!! Welcome to TASS Movidius Facenet Classifier, please wait while the program initiates !!

-- Running on Python 3.5.2 (default, Nov 23 2017, 16:37:01)
[GCC 5.4.0 20160609]

-- Imported Required Modules
-- Movidius Connected
-- Allocated Graph OK
-- Initiating JumpWayMQTT Device
-- JumpWayMQTT Device Initiated
-- JumpWayMQTT Device Connection Initiating
-- JumpWayMQTT Device Connection Initiated
-- IoT JumpWay Initiated

-- Classifier Initiated

-- FACENET TEST MODE STARTING
-- STARTED: 2018-08-23 14:42:31.833471

-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status
-- Total Difference is: 1.76318955421
-- MATCH 020.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.81890833378
-- MATCH 008.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.88612890244
-- MATCH 012.jpg

-- Published to Device Warnings Channel
-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status
-- Total Difference is: 1.81377631426
-- MATCH 016.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.878862679
-- MATCH 015.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.78870594501
-- MATCH 019.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.39330101013
-- MATCH 004.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.7547147274
-- MATCH 102.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.86856722832
-- MATCH 011.jpg

-- Published to Device Warnings Channel
-- Total Difference is: 1.37876737118
-- MATCH 003.jpg

-- Published to Device Warnings Channel
-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status
-- Total Difference is: 1.83961397409
-- MATCH 007.jpg

-- Published to Device Warnings Channel

-- FACENET TEST MODE ENDING
-- ENDED: 2018-08-23 14:42:35.234321
-- TESTED: 11
-- IDENTIFIED: 11
-- TIME(secs): 3.400850772857666

!! SHUTTING DOWN !!**

As you can see the program simply runs through the various frames / images of the testing folder indicating a generic match that does not specify however with what is matching

The other problem I found is about the WebCam (WebCam.py) where, after launching the program, no inference is shown on the console but simply an indication that the stream is running
To perform one, pass me the term, "barbaric" inference and / or recognition I went to contact the indicated server via browser (http: // myserverid: 8080 / index.html) and I noticed an interesting thing: this operation causes a crash on both the program and the site but it returns me back to the main folder (TASS-Facenet) a screenshot ONLY IF A RECOGNITION WAS SUCCESSFUL of the frame that the camera was showing. Even here I have not found a way to get a better graphical interface

**!! Welcome to TASS Facenet WebCam Classifier, please wait while the program initiates !!

-- Running on Python 3.5.2 (default, Nov 23 2017, 16:37:01)
[GCC 5.4.0 20160609]

-- Movidius Connected
-- Allocated Graph OK
-- Initiating JumpWayMQTT Device
-- JumpWayMQTT Device Initiated
-- JumpWayMQTT Device Connection Initiating
-- JumpWayMQTT Device Connection Initiated
-- IoT JumpWay Initiated
-- JumpWayMQTT Device Connected
rc: 5

-- Classifier Initiated

-- Published to Device Status
-- CONNECTED TO WEBCAM
-- Server started on 172.20.1.12:8080
-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status
-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status
-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status
-- JumpWayMQTT Device Connected
rc: 5
-- Published to Device Status**

As you can see there is no recognition, match or inference .... simply it is indicated that the stream of the camera is working (even the camera itself tells me via led that it's correct functioning). In your guide the result to be obtained should be the like the following but I can not get this kind of output:

-- Saved frame
-- Total Difference is: 1.0537698864936829
-- MATCH
-- Published: 30
-- Published to Device Warnings Channel

Mindful of the fact that I probably made mistakes or did not understand the basic steps, I wish you a good day and a good job

Thanks for the help and support

Prati Mattia