[Required] Your device (RaspberryPi3, LaptopPC, or other device name): Raspberry PI3

[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name): armv7l

[Required] Your OS (Raspbian, Ubuntu1604, or other os name): Raspbian

[Required] Details of the work you did before the problem occurred:


Finish all steps described in the article.


[Required] Error message:


ImportError: No module named 'mvnc'


[Required] Overview of problems and questions:


Does OpenVino include mvnc?

Dear Master Pinto,

in your very thorough and kind step by step I miss the NCSDK v2 setup... when do you consider best to install it (after which step number)?

After seeing some stretch issues, do you use ubuntu 16.04 in rasp? which exactly (@ rasp site there are two: ubuntu mate & snappy)

thx!!

Hello,PINTO0309!
When I trained MobileNet-SSD on my own datasets(20 different classes, 400 images per class), I found the results of FP32 and FP16 be completely different(the caffe model test detection_eval is 0.91). And the same problem also happend when I trained YOLOv3 on my own datasets(follow your other github program).
Are you trying to train the model with your own data? I doubted if the openvino and ncsdk can only convert the specific model perfectly. Could you give me some advices? Thank you!

Hi @PINTO0309
This is such an amazing work! May I link to it from librealsense/examples#community-projects?

Hi, I tried to load multi NCS2 on RaspberryPi3B+ and Ubuntu16.04LTS with intel i7-7700k just like what you do, using the IEPlugin function. I actually plugged 2 NCS2, and called the function for 2 times, but it failed with a error message saying NO_DEVICE when loading the second NCS2. Have you encountered the same problem before? I thought a plugin is an abstract for a device, am I right?

Dear PINTO,

I noted that when doing "make examples -j1" when installing ncsdk it automatically installs OpenCV...
I understand that if you are doing a fresh install there is no need to install it to later uninstall it and then install it again in your step-by-step, so do I really need to make the examples in this step or I can psotpone this step after the whole step-by-step process?

I also see now there are repeated steps regarding the installation of protobuf 3.5.1, do I really need to install it twice?

Thanks in avdance!

btw: you made an amazing work! thks a lot!

i have 2 NCS2
i want to USE 1-NCS2 to run first model and use another NCS2 to run second model,it is possible?
how do i set?

I have suffer from memory leak when running MultiStickSSDwithRealSense_OpenVINO_NCS2.py with a single stick and USB Webcam.

The memory leak occurs on Laptop and Odroid XU4 with rate about 200MB/minutes.

I can run SingleStickSSDwithUSBCamera_OpenVINO_NCS2.py normally without any memory leak, my guess is the problem is at the multiprocessing where the already processed image didn't removed from the memory.

Laptop Specification:
Ubuntu 18.04.1 (x64)
i7-4710HQ (2.5 GHz / 4 Cores)
GTX 850M (2 GB VRAM)
8 GB RAM

ODROID XU4 Specification:
Ubuntu Mate 16.04 (ARM)
Samsung Exynos5422 Cortex™-A15 2Ghz and Cortex™-A7 (2 GHz / 8 Cores)
Mali-T628 MP6
2 GB RAM

Camera:
Laptop Webcam, Logitech C270, PS3 EYE

Could you share the way of convert MobileNet-ssd to IR?
I convert Mobilenet-ssd v2 to IR, but got lots of error. Here is my approach:

python3 .\mo_tf.py `
--input_model E:\ssdlite_mobilenet_v2\frozen_inference_graph.pb `
--model_name ssd_mobilenet_v2 `
--output_dir E:\ssdlite_mobilenet_v2\ `
--data_type FP16 `
--batch 1 `
--tensorflow_object_detection_api_pipeline_config E:\ssdlite_mobilenet_v2\pipeline.config `
--tensorflow_use_custom_operations_config .\extensions\front\tf\ssd_v2_support.json

There are many errors when I use this converted IR model.

Dear "demigod" Pinto,

We recently implemented a multiprocessing version in our own way, pretty similar to yours, but we got some stability issues. In fact we were somehow surprised by the lack of multiprocessing examples in zoo collection and just some multithreading ones for multistick, being the later a bottle neck due to GIL, so no big advantages from multistick at all.

Our approach was based in API v1 with multiprocessing in fork (default) mode. why did you choose forkserver + daemon?

Which is the actual increase in detection fps (not screen rendering) that you achieved by each extra stick added? from 1 to 2, then 3 and finally 4

One final question about performance. the
allocate_with_fifos command defaults to 2 elements per each input / output fifo. does it ensure that stick will be all time busy without waiting times? i mean if it allows you to push another item to the graph queue while the previous one is being processed.

Anyway very excellent job on your side!

[Required] Your device (RaspberryPi3):

[Required] Your device's CPU architecture (armv7l):

[Required] Your OS (Raspbian):

[Required] Details of the work you did before the problem occurred:

I followed your steps to install ncsdk on the Raspberry Pi. An error occurred while executing make examples
[Required] Error message:

Network Input tensors ['data#24']

Network Output tensors ['prob#38']

Blob generated
W: [ 0] ncDeviceOpen:528 ncDeviceOpen() XLinkBootRemote returned error 3

[Error 7] Toolkit Error: USB Failure. Code: Error opening device
Makefile:64: recipe for target 'profile' failed
make[3]: *** [profile] Error 255
make[3]: Leaving directory '/home/pi/ncsdk/examples/caffe/AlexNet'
Makefile:12: recipe for target 'AlexNet' failed
make[2]: *** [AlexNet] Error 2
make[2]: Leaving directory '/home/pi/ncsdk/examples/caffe'
Makefile:12: recipe for target 'caffe/.' failed
make[1]: *** [caffe/.] Error 2
make[1]: Leaving directory '/home/pi/ncsdk/examples'
Makefile:57: recipe for target 'examples' failed
make: *** [examples] Error 2
[Required] Overview of problems and questions:
Unable to find device

[Required] Your device (RaspberryPi3, LaptopPC, or other device name):
RaspberryPi3 b+ , NCS2 x 4 ,
[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name):
armv7l
[Required] Your OS (Raspbian, Ubuntu1604, or other os name):
Rasbian
[Required] Details of the work you did before the problem occurred:

I have four NCS2 and I`m trying to run different models on each neural stick independently.

For example, first neural stick run face detection, next one run emotion recognition, and third one run image classificaiton ..
Is it possible ? I saw your MutiModel(FaceDetection, EmotionRecognition) which is merged
model but what I want to do is what I said above.


I`m really appreciate your project
Thanks

So first off thanks again for the tuned performance. I verified 12 FPS peak on the Raspberry Pi with the Realsense D435.

The thing I noticed (see videos below), is that the scaling/positioning math behind the bounding boxes after 97e92fa (here) seem to be off.

So prior to this commit, they follow me left/right in the live video feed correctly (my center, and the center of the bounding box match well), and the scale matches well too (i.e. my width/height, and the bounding box width/height match well).

After the commit, I notice the following:

1. The positioning is off: when object is on the left side, the position of the bounding box is too far to the left. When the object is on the right, the center of the bounding box is too far to the right, etc.
2. The left/right size is also off. The box is too large at least in the left/right dimension (and probably also up/down, it seems, but less tested).

Here's a before/correct video (from checkout 80a564c):
https://photos.app.goo.gl/H5ta1X7KnqTdsz7j7

Notice that the bounding box tracks my center well and also my size well.

And here's the after/incorrect video (from checkout 055636a):
https://photos.app.goo.gl/6zfMgGKaYjbrZRQZ9

Notice in this case the box is too wide (and also seemingly too tall), and goes to far left/right when the object (me) is located left/right of the video feed.

Thanks again! I'll also try to hunt to see if I can find where this scale/position error was introduced.

[Required] Your device: Surface book 2 15 inch

[Required] Your device's CPU architecture: i7 8650U

[Required] Your OS: Windows 10
[Required] Details of the work you did before the problem occurred:

I was doing a benchmarking between CPU, GPU, and NCS2. I found the speed: CPU > MYRAID > GPU. It seems the result is not reasonable.

        if device == "CPU":
            model_xml = 'model_ir/fp32/frozen_alexnet_model.xml'
            model_bin = os.path.splitext(model_xml)[0] + ".bin"
            plugin = IEPlugin(device='CPU') # 使用CPU运行模型，要用NCS的话改为MYRIAD
            net = IENetwork(model=model_xml, weights=model_bin)  # 用NCS的话也需要更改
        elif device == "GPU":
            model_xml = 'model_ir/fp32/frozen_alexnet_model.xml'
            model_bin = os.path.splitext(model_xml)[0] + ".bin"
            plugin = IEPlugin(device='GPU')
            net = IENetwork(model=model_xml, weights=model_bin)
        else:
            model_xml = 'model_ir/fp16/frozen_alexnet_model.xml'
            model_bin = os.path.splitext(model_xml)[0] + ".bin"
            plugin = IEPlugin(device='MYRIAD')
            net = IENetwork(model=model_xml, weights=model_bin)

        net.batch_size = batch
        input_blob = next(iter(net.inputs))
        exec_net = plugin.load(network=net)

        print('tick tok...')
        start_time = time.time()

        for images, images_path in get_batches_fn(batch):
            outputs = exec_net.infer(inputs={input_blob: images})

            for indx in range(batch):
                class_name = class_names[np.argmax(outputs['prob'][indx])]
                probs = outputs['prob'][indx, np.argmax(outputs['prob'][indx])]

[Required] Overview of problems and questions:
Why GPU does not have a better performance? Is it normal?

Hello,

Many thanks for your code.

Got about 15 FPS, using USB Webcam on Rpi 3 B+ and 2 ncs2.
Have you bench using 4 ncs on rpi ?

I use https://github.com/umlaeute/v4l2loopback to feed trafic cam view.

I get better result starting cam in 640x480 and cropping like this:

color_image=cv2.resize(color_image,(532,400))
color_image=color_image[100:100+300,116:116+300]

I also try to use the script with yolov3 but got buffer overflow.

Regards,

Pierre

I've been trying to use this repository to run a custom MobileNet-SSD model trained with tensorflow and converted to IR. I have validated a FP32 version of this model using OpenVino's object_detection_sample_ssd, but I'm using a FP16 version for this repository.

I'm running SingleStickSSDwithUSBCamera_OpenVINO_NCS2.py, modified to load my .xml/.bin model and my labels, but I never get any detections. Any idea why? Thanks!

pyrealsense2を用いたプログラムをsystemdを用いて起動を試みたのですが(python3)、以下のエラーにより失敗します。

ImportError: No module named 'pyrealsense2'

systemdを用いずに起動した際は、"ImportError"は出ず、プログラムは正常に動作します。
また、パッケージのバージョンは全てreadmeに従っています。

宜しくお願い致します。

English
I attempted to run a program using pyrealsense2 with systemd (python 3), but it fails with the following error.

ImportError: No module named 'pyrealsense2'

When run the program without using systemd, it works normally.
Also, all package versions compliant with readme.

Thank you.

repeated. sorry!

Anyway very excellent job on your side!

[Required] Your device (RaspberryPi3, LaptopPC, or other device name): RaspberryPi4

[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name): aarch64

[Required] Your OS (Raspbian, Ubuntu1604, or other os name): Ubuntu 18.04 LTS

[Required] Details of the work you did before the problem occurred:
I have followed all the steps in your guide and I keep arriving at the same error. I have also tried to manually install opencv myself but still cant seem to figure out why the code doesnt work. I can run sample object detection codes on single images but not videos. This is the error i get when I try to run SingleStickSSDwithRealSense_OpenVINO_NCS2.py






[Required] Error message:
Traceback (most recent call last):
File "SingleStickSSDwithRealSense_OpenVINO_NCS2.py", line 33, in
net = cv2.dnn.readNet('lrmodel/MobileNetSSD/MobileNetSSD_deploy.xml', 'lrmodel/MobileNetSSD/MobileNetSSD_deploy.bin')
cv2.error: OpenCV(4.3.0) /home/ubuntu/opencv/modules/dnn/src/dnn.cpp:3541: error: (-2:Unspecified error) Build OpenCV with Inference Engine to enable loading models from Model Optimizer. in function 'readFromModelOptimizer'






[Required] Overview of problems and questions:
At first the script would not be able to import cv2 at all but after manually installing OpenCV 4.3.0 I was able to import cv2 and run the dnn sample code.
I am very new to this so I would appreciate all the help I can get.

Hi, PINTO0309
I have the environment: Raspberry Pi + Raspbian Stretch+ RealSense D435.
There are some problems with my D435 connecting to my Pi.
The screenshot as follows:

1. I have no RGB module, only stereo module in my $ realsense-viewer
2. Incomplete frame (38%) and no frames received. (Maybe the transfer speed is not enough with usb2.0? But it is perfect in your demo. )

Why the position displayed by the detection frame is not the position of the instant picture, he will probably be slower by 3~5 frames.

I'm in learning to use openvino and movidius1
your github example is great!!

And if you pleased, i would like to know how you merge object detection and face detection before convert to IR

Thank you

Hi PINTO0309,

Thanks for all your work here and particularly for documenting it all. I've bene able to successfully follow everything, which I think speaks to your thoroughness. :-)

So I've setup and run, with RealSense D435, the following:

1. No NCS with TensorFlow (here
2. NCS1 (this repository, with NCSDK)
3. NCS2 (this repository, with OpenVINO), using python3 MultiStickSSDwithRealSense_OpenVINO_NCS2.py - but I only have 1x NCS2 device.

So I'm seeing the following framerates:

1. 1 FPS - No NCS
2. 6 FPS - 1x NCS1
3. 5 FPS - 1x NCS2

So I'm curious to see that the NCS2 with OpenVINO actually seems to be slightly slower than the NCS1 with NCSDK.

Now my question is, it feels like I'm running the wrong command for 3... am I?

I'm running python3 MultiStickSSDwithRealSense_OpenVINO_NCS2.py and the only other options I see there are:

• MultiStickSSDwithRealSense.py
• SingleStickSSDwithRealSense.py
• SingleStickSSDwithUSBCamera_OpenVINO_NCS2.py

So it kind of makes me think that SingleStickSSDwithRealSense_OpenVINO_NCS2.py, or something like that, is missing from github (like forgot to be committed, or similar).

Thoughts?

Oh and to see if SingleStickSSDwithRealSense.py happened to be compatible with both NCS1 (NCSDK) and NCS2 (OpenVINO), I rant it with the NCS2 connected (and no NCS1 connected), and got:

python3 SingleStickSSDwithRealSense.py 
Traceback (most recent call last):
  File "SingleStickSSDwithRealSense.py", line 13, in <module>
    from mvnc import mvncapi as mvnc
ImportError: No module named 'mvnc'

Thanks again!
-Brandon

[Required] Your device (RaspberryPi3, LaptopPC, or other device name): LaptopPC

[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name): x64

[Required] Your OS (Raspbian, Ubuntu1604, or other os name): Ubuntu 1804

[Required] Details of the work you did before the problem occurred:

I have run the SingleStickSSDwithRealSense_OpenVINO_NCS2.py code.

Components used: Intel RealSense camera D435, Intel NCS2 and LaptopPC


[Required] Error message:






[Required] Overview of problems and questions:

I want to include(train) all the 80 classes of coco dataset and use it in the Single StickSSDwithRealSense_OpenVINO_NCS2 python program

Hi there !!
I have SSD mobile net model (written in tensorflow) for image classification..... and it works well on my windows pc with a dlib tracker added to it..... but i need to use a raspberry pi for prototyping ...
i install tensorflow on my RPi and try to execute the file ( on the python 3 IDLE) and it shows me nothing but no error is generated.... could you tell me my mistake ?

So this could be a synchronous/asynchronous observer phenomenon, I'm not sure, and will investigate more. That said, I did want to drop in a comparison here of NCS1 (synchronous) vs. NCS2 (asynchronous):

NCS1 (synchronous): https://photos.app.goo.gl/ejezgpvmMz7U53Qu9
NCS2 (asynchronous): https://photos.app.goo.gl/BExZNpWVaTvLA3wn8

Again, I need to check whether this is just an observational thing of synchronous/asynchronous, but wanted to share these in the meantime in case they give an 'ah-ha' moment. :-)

Thanks again,
Brandon

Great work, thank you very much!

In this gist https://gist.github.com/treinberger/c63cb84979a4b3fb9b13a2d290482f4e I ported your code to get the camera input from the PiCam, but I discovered that the framerate is really low (about 1FPS). Have you got any ideas why this could be the case? I ported your code because I currently don't have a usb webcam at hand. My environment is Raspberry Pi 3+, some wideangle camera and the NCS2.

LaptopPc+ x86_64+ Ubuntu1604 ssd_300_vgg
use the program from https://github.com/pierluigiferrari/ssd_keras

how to convert the .pb model to IR?
l use
python3 mo_tf.py --data_type FP16 --input_shape [1,300,300,3] --input_model /path/to/my/pbmodel
it's wrong.
[ ERROR ] Shape [ 1 3 -1 256] is not fully defined for output 0 of "conv8_2/convolution". Use --input_shape with positive integers to override model input shapes. [ ERROR ] Cannot infer shapes or values for node "conv8_2/convolution". [ ERROR ] Not all output shapes were inferred or fully defined for node "conv8_2/convolution". For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #40. [ ERROR ] [ ERROR ] It can happen due to bug in custom shape infer function <function Convolution.infer at 0x7fe94897d378>. [ ERROR ] Or because the node inputs have incorrect values/shapes. [ ERROR ] Or because input shapes are incorrect (embedded to the model or passed via --input_shape). [ ERROR ] Run Model Optimizer with --log_level=DEBUG for more information. [ ERROR ] Stopped shape/value propagation at "conv8_2/convolution" node. For more information please refer to Model Optimizer FAQ (<INSTALL_DIR>/deployment_tools/documentation/docs/MO_FAQ.html), question #38.
how to correct convert? need other parameter? and parameter's content?

And what should l change the code if l use your code to detect.

[Required] Your device (RaspberryPi3, LaptopPC, or other device name): Intel NUC5i7RYHR

[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name): x86_64

[Required] Your OS (Raspbian, Ubuntu1604, or other os name): Ubuntu1604

[Required] Details of the work you did before the problem occurred:
I'm following your https://github.com/PINTO0309/MobileNet-SSD-RealSense/README.md for an Intel 5th gen NUC running Ubuntu 16.04.6 LTS 64bit.

Specifically, I have an NCS 1 so I'm using the NCSDK2 and following the instruction under heading:
"Work with RaspberryPi3 (or PC + Ubuntu16.04 / RaspberryPi + Ubuntu Mate)"

At step 7, I found your TBBonamd64x86x64 repository and downloaded: https://github.com/PINTO0309/TBBonamd64x86x64/raw/master/libtbb-dev_2018U2_amd64.deb

I'm then trying to build OpenCV 3.4.2 from source (step 9.1) but get stuck at the compilation stage with an error.

[Required] Error message:
/usr/local/include/tbb/atomic.h:222:34: error: ‘tbb::internal::atomic_impl::my_storage’ has incomplete type
aligned_storage<T,sizeof(T)> my_storage;
^
/usr/local/include/tbb/atomic.h:95:8: note: declaration of ‘struct tbb::internal::aligned_storage<long int, 8ul>’
struct aligned_storage;
^
/usr/local/include/tbb/atomic.h: In instantiation of ‘struct tbb::internal::atomic_impl’:
/usr/local/include/tbb/atomic.h:338:8: required from ‘struct tbb::internal::atomic_impl_with_arithmetic<long unsigned int, long unsigned int, char>’
/usr/local/include/tbb/atomic.h:444:1: required from here

[Required] Overview of problems and questions:
Please help!

[Required] Your device (RaspberryPi3, LaptopPC, or other device name):
Raspberry Pi3b / Raspberry pi3b+
[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name):

[Required] Your OS (Raspbian, Ubuntu1604, or other os name):
Raspbian / Ubuntu 16.04
[Required] Details of the work you did before the problem occurred:

i've already tried your project PINTO0309/MobileNet-SSDLite-RealSense-TF . Thank you for your project. But for my project, that was slow and unstable. So now,I'm considering using NCS v1 or v2 to improve performance of raspberry pi3b/3b+ with realsense.




[Required] Error message:






[Required] Overview of problems and questions:

I wonder how much fps or perfermance difference exists between no NCS and using NCS v1 and NCS v2. Thank you.

Device: Raspberry Pi 3 B+

CPU Arch.: armv7l

OS: Raspbian

I have trained a custom 1 class mobile net ssd v2 network using Tensorflow Object Detection API and successfully converted to to IR model using OpenVino 2019 R1.1. Then I substituted the model files in the MultiStickWithPiCamera.py example with my custom model. The predict_async thread raises an error.

Error message:

line 174, in predict_async
cnt, dev = heapq.heappop(self.heap_request)
IndexError: index out of range

[Required] Your device (RaspberryPi3, LaptopPC, or other device name):
Rpi4
[Required] Your device's CPU architecture (armv7l, x86_64, or other architecture name):
armv7l
[Required] Your OS (Raspbian, Ubuntu1604, or other os name):
Raspbian
[Required] Details of the work you did before the problem occurred:
Crashes after running inference 10 min or so in raspberry pi 4
"MultiStickSSDwithPiCamera_OpenVINO_NCS2.py "
[Required] Error message:
E: [xLink] [ 262150] [EventRead00Thr] eventReader:218 eventReader thread stopped (err -4)
E: [xLink] [ 262150] [Scheduler00Thr] eventSchedulerRun:576 Dispatcher received NULL event!
E: [global] [ 262151] [python3] XLinkReadDataWithTimeOut:1494 Event data is invalid
E: [ncAPI] [ 262151] [python3] ncFifoReadElem:3510 Packet reading is failed.
E: [watchdog] [ 262340] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 263339] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 264338] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 265337] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 266336] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 267336] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 268335] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 269334] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 270333] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 271332] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 272331] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 273331] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 274331] [WatchdogThread] sendPingMessage:121 Failed send ping message: X_LINK_ERROR
E: [watchdog] [ 274332] [WatchdogThread] watchdog_routine:315 [0x30a02e0] device, not respond, removing from watchdog

E: [xLink] [ 274676] [WatchdogThread] dispatcherWaitEventComplete:774 waiting is timeout, sending reset remote event

[Required] Overview of problems and questions:
Crashes after running inference 10 min or so in raspberry pi 4
"MultiStickSSDwithPiCamera_OpenVINO_NCS2.py "

It occurs an error when I using the following commands with my raspberry pi:
$ sudo apt install -y git libssl-dev libusb-1.0-0-dev pkg-config libgtk-3-dev libglfw3-dev at-spi2-core libdrm*
###########################################
Reading package lists... Done
Building dependency tree
Reading state information... Done
Note, selecting 'libdrm-radeon1-dbg' for glob 'libdrm*'
Note, selecting 'libdrm-freedreno1' for glob 'libdrm*'
Note, selecting 'libdrm-nouveau1' for glob 'libdrm*'
Note, selecting 'libdrm-nouveau2' for glob 'libdrm*'
Note, selecting 'libdrm2-dbg' for glob 'libdrm*'
Note, selecting 'libdrm-nouveau1a' for glob 'libdrm*'
Note, selecting 'libdrm-amdgpu1' for glob 'libdrm*'
Note, selecting 'libdrm-omap1' for glob 'libdrm*'
Note, selecting 'libdrm-omap1-dbg' for glob 'libdrm*'
Note, selecting 'libdrm2' for glob 'libdrm*'
Note, selecting 'libdrm-nouveau1a-dbg' for glob 'libdrm*'
Note, selecting 'libdrmaa1.0' for glob 'libdrm*'
Note, selecting 'libdrmaa-dev' for glob 'libdrm*'
Note, selecting 'libdrm-radeon1' for glob 'libdrm*'
Note, selecting 'libdrm-dev' for glob 'libdrm*'
Note, selecting 'libdrm-nouveau1-dbg' for glob 'libdrm*'
E: Unable to locate package libglfw3-dev
###########################################

what's the problems?