Fixes for wine detection tutorial #13886

larroy · 2019-01-15T15:33:46Z

Description

see tittle

Checklist

Essentials

Please feel free to remove inapplicable items for your PR.

The PR title starts with [MXNET-$JIRA_ID], where $JIRA_ID refers to the relevant JIRA issue created (except PRs with tiny changes)
Changes are complete (i.e. I finished coding on this PR)
All changes have test coverage:
Unit tests are added for small changes to verify correctness (e.g. adding a new operator)
Nightly tests are added for complicated/long-running ones (e.g. changing distributed kvstore)
Build tests will be added for build configuration changes (e.g. adding a new build option with NCCL)
Code is well-documented:
For user-facing API changes, API doc string has been updated.
For new C++ functions in header files, their functionalities and arguments are documented.
For new examples, README.md is added to explain the what the example does, the source of the dataset, expected performance on test set and reference to the original paper if applicable
Check the API doc at http://mxnet-ci-doc.s3-accelerate.dualstack.amazonaws.com/PR-$PR_ID/$BUILD_ID/index.html
To the my best knowledge, examples are either not affected by this change, or have been fixed to be compatible with this change

stu1130 · 2019-01-15T17:13:12Z

@mxnet-label-bot add [pr-awaiting-review]
Thanks for your great work @larroy

larroy · 2019-01-15T18:54:42Z

#9094

jlcontreras

LGTM

vandanavk · 2019-02-05T20:01:41Z

@aaronmarkham for review/merge

aaronmarkham

I'm not going to hold this up with my testing as my hardware is Pi Model B v2 and only has 512MB of RAM. I keep getting stuck trying into install numpy. The changes look fine, but at the moment I can't validate the functionality of the tutorial.
One thing to note though - the instructions tell you to follow the install page's setup, but those tell you to create a Python 3 env, whereas this tutorial is 2.7. Minor detail, but maybe this could be fixed.

Here's the flow I started (but stopped at the install step as it is hanging):
Format your SD card with SD Card Formatter.
Download, unzip, then copy contents of the NOOBS (network install) zip file to SD card.
Boot up the Pi and follow the prompts to install the headless Raspbian (smaller and might help with memory issues).
Login to the pi with user pi and password raspberry.
Install pip with:

sudo apt-get install python-pip

Install virtualenv with:

sudo pip install virtualenv

Create a Python 2.7 environment for MXNet with:

virtualenv -p `which python` mxnet_py27

Download the mxnet wheel with:

wget https://mxnet-public.s3.amazonaws.com/install/raspbian/mxnet-1.5.0-py2.py3-none-any.whl

Install mxnet with:

pip install mxnet-1.5.0-py2.py3-none-any.whl

That wheel is something I created. It would be nice if we hosted some tagged versions somewhere, so people can grab a wheel with v1.3.1 or the latest RC.
Also, I don't see that this tutorial has a link from the tutorial's index page. So how does one find this tutorial?

aaronmarkham

I still haven't seen this work yet, but I can tell you that the earlier step where you copy the model params file is incorrect. The file listed isn't in the tar. Also the dependencies setup isn't complete.

I have my notes on how to get MXNet running on a Pi 3 with Stretch here:
https://gist.github.com/aaronmarkham/b3d0c715abca92ccc4b79ed0f19d1916

A Pi 2 doesn't work with MXNet (using the cross-compiled wheel) - it core dumps.

One further issue with this tutorial is that it requires OpenCV2 to be installed separately. This would require a build from source as there doesn't seem to be a package available for Stretch. Then I ran out of disk space on my 8GB card before I could go further. This could be avoided though since the image processing steps are already part of MXNet's port of OpenCV operators. We don't need OpenCV 2 here.

I'm not going to approve or request changes. If you know this update works on your test devices that say so, and we'll merge this update. However, I think that this tutorial needs to be modernized to use the latest MXNet features for handling images. And we should also have the Install page updated so that it has the appropriate dependencies.

Additionally, I found it a pain to have to compile on Ubuntu, and I really wish that we offered wheels prepped for whatever Pi & MXNet versions we can support.

aaronmarkham · 2019-02-14T23:13:20Z

I put the edits for MXNet installation on Raspberry Pi in this PR: #14172

aaronmarkham · 2019-02-14T23:14:24Z

I also just fixed the conflict from the new license header.

ankkhedia

LGTM

ankkhedia · 2019-02-15T22:40:06Z

@larroy Could you look into comments by @aaronmarkham

aaronmarkham · 2019-02-25T19:00:25Z

I updated the Pi instructions here: #14172
So @larroy, if you know the tutorial works, then LMK and I'll merge this update. I can raise my issues with the preprocessing part of the tutorial in a separate issue.

szha · 2019-02-25T20:58:17Z

docs/tutorials/embedded/wine_detector.md

@@ -100,13 +101,14 @@ with open('synset.txt', 'r') as f:
    synsets = [l.rstrip() for l in f]

 # Load the network parameters
-sym, arg_params, aux_params = mx.model.load_checkpoint('Inception_BN', 0)
+sym, arg_params, aux_params = mx.model.load_checkpoint('Inception-BN', 126)


magic number?

It is the number of epoch.

Right, what I mean is this probably shouldn't be hard-coded?

Maybe, but if you follow the instructions, the model donloaded has that epoch for the checkpoint. What do you suggest instead?

vandanavk · 2019-03-04T21:53:48Z

@mxnet-label-bot update [Example, pr-awaiting-response]

karan6181 · 2019-03-17T15:01:21Z

@larroy Could you please address the review comments? Thanks!

wkcn

Thank you for the fix! LGTM

abhinavs95 · 2019-03-29T21:59:32Z

@larroy Could you please give an update on the PR? Thanks

piyushghai · 2019-04-09T00:45:36Z

@larroy Gentle ping...

Roshrini · 2019-04-17T15:34:49Z

@aaronmarkham @larroy Is this PR good to go?

aaronmarkham · 2019-04-17T15:55:39Z

@larroy has an unresolved comment here. Once that's done and he's sure it works, I'm ok with the PR.

larroy · 2019-04-22T19:28:37Z

@aaronmarkham I responded to the comment.

larroy · 2019-04-22T19:28:53Z

@mxnet-label-bot remove [pr-awaiting-response]

larroy · 2019-04-22T19:29:12Z

@mxnet-label-bot add [pr-awaiting-review]

wkcn

LGTM. Thank you!

larroy requested a review from szha as a code owner January 15, 2019 15:33

marcoabreu added the pr-awaiting-review PR is waiting for code review label Jan 15, 2019

larroy mentioned this pull request Jan 15, 2019

Segmentation Fault #9507

Closed

Fixes for wine detection tutorial

eca332c

larroy force-pushed the wine_detector branch from ca976ab to eca332c Compare January 16, 2019 13:43

jlcontreras approved these changes Jan 18, 2019

View reviewed changes

aaronmarkham reviewed Feb 7, 2019

View reviewed changes

aaronmarkham mentioned this pull request Feb 12, 2019

raspberry pi broken installation path #14135

Closed

aaronmarkham reviewed Feb 14, 2019

View reviewed changes

Merge branch 'master' into wine_detector

7dc9a2e

ankkhedia approved these changes Feb 15, 2019

View reviewed changes

szha reviewed Feb 25, 2019

View reviewed changes

marcoabreu added Example pr-awaiting-response PR is reviewed and waiting for contributor to respond and removed pr-awaiting-review PR is waiting for code review labels Mar 4, 2019

wkcn approved these changes Mar 18, 2019

View reviewed changes

marcoabreu removed the pr-awaiting-response PR is reviewed and waiting for contributor to respond label Apr 22, 2019

marcoabreu added the pr-awaiting-review PR is waiting for code review label Apr 22, 2019

wkcn approved these changes Apr 28, 2019

View reviewed changes

aaronmarkham merged commit 1af29e9 into apache:master Apr 29, 2019

access2rohit pushed a commit to access2rohit/incubator-mxnet that referenced this pull request May 14, 2019

Fixes for wine detection tutorial (apache#13886)

9012bab

haohuanw pushed a commit to haohuanw/incubator-mxnet that referenced this pull request Jun 23, 2019

Fixes for wine detection tutorial (apache#13886)

35069d5

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fixes for wine detection tutorial #13886

Fixes for wine detection tutorial #13886

larroy commented Jan 15, 2019

stu1130 commented Jan 15, 2019

larroy commented Jan 15, 2019

jlcontreras left a comment

vandanavk commented Feb 5, 2019

aaronmarkham left a comment •

edited

Loading

aaronmarkham left a comment

aaronmarkham commented Feb 14, 2019

aaronmarkham commented Feb 14, 2019

ankkhedia left a comment

ankkhedia commented Feb 15, 2019

aaronmarkham commented Feb 25, 2019

szha Feb 25, 2019

wkcn Mar 18, 2019

szha Mar 18, 2019

wkcn Mar 18, 2019

larroy Apr 19, 2019 •

edited

Loading

vandanavk commented Mar 4, 2019

karan6181 commented Mar 17, 2019

wkcn left a comment •

edited

Loading

abhinavs95 commented Mar 29, 2019

piyushghai commented Apr 9, 2019

Roshrini commented Apr 17, 2019

aaronmarkham commented Apr 17, 2019

larroy commented Apr 22, 2019

larroy commented Apr 22, 2019

larroy commented Apr 22, 2019

wkcn left a comment

Fixes for wine detection tutorial #13886

Fixes for wine detection tutorial #13886

Conversation

larroy commented Jan 15, 2019

Description

Checklist

Essentials

stu1130 commented Jan 15, 2019

larroy commented Jan 15, 2019

jlcontreras left a comment

Choose a reason for hiding this comment

vandanavk commented Feb 5, 2019

aaronmarkham left a comment • edited Loading

Choose a reason for hiding this comment

aaronmarkham left a comment

Choose a reason for hiding this comment

aaronmarkham commented Feb 14, 2019

aaronmarkham commented Feb 14, 2019

ankkhedia left a comment

Choose a reason for hiding this comment

ankkhedia commented Feb 15, 2019

aaronmarkham commented Feb 25, 2019

szha Feb 25, 2019

Choose a reason for hiding this comment

wkcn Mar 18, 2019

Choose a reason for hiding this comment

szha Mar 18, 2019

Choose a reason for hiding this comment

wkcn Mar 18, 2019

Choose a reason for hiding this comment

larroy Apr 19, 2019 • edited Loading

Choose a reason for hiding this comment

vandanavk commented Mar 4, 2019

karan6181 commented Mar 17, 2019

wkcn left a comment • edited Loading

Choose a reason for hiding this comment

abhinavs95 commented Mar 29, 2019

piyushghai commented Apr 9, 2019

Roshrini commented Apr 17, 2019

aaronmarkham commented Apr 17, 2019

larroy commented Apr 22, 2019

larroy commented Apr 22, 2019

larroy commented Apr 22, 2019

wkcn left a comment

Choose a reason for hiding this comment

aaronmarkham left a comment •

edited

Loading

larroy Apr 19, 2019 •

edited

Loading

wkcn left a comment •

edited

Loading