profile
viewpoint
If you are wondering where the data of this site comes from, please visit https://api.github.com/users/fharper/events. GitMemory does not store any data, but only uses NGINX to cache data for a period of time. The idea behind GitMemory is simply to give users a better reading experience.

fharper/macsetup 39

Scripts to setup my mac

fharper/devrel-interviews 13

Reverse Interview Questions for Developer Advocate/Technical Evangelist roles

fharper/cantdeletemyaccount 3

A public shaming list of websites not having a "delete my account" option

fharper/rain 3

Simple macOS menubar app that play rainforest sound

fharper/FoxTheMilk 2

A Remember the Milk Firefox OS application

fharper/jekyllExternalLinksPlugin 2

A Jekyll plugin that adds a target blank to all external links in posts and pages.

fharper/speakerkit 2

My public speaking kit for events organizers

fharper/mytalks 1

List of my previous & upcoming public speaking engagements

fharper/alfred-npmjs 0

Alfred 3 workflow to search for npm packages

fharper/ama 0

Ask Me Anything

startedTalhaMangarah/tiktok-downloader-ios-shortcut

started time in 17 hours

issue openedmindee/devrel

Write down why we made the switch to WordPress

Write down why we made the switch to WordPress in this repository. That could also be a blog post!

created time in 2 days

issue openedmindee/devrel

Fix broken links in the blog

Fix broken links in the blog: mostly links pointing to the documentation since no forward was done when the documentation structure changed.

created time in 2 days

issue openedmindee/devrel

Create a Slack campaign

Create a Slack campaign so we can get our actual users to join it since the first newsletter didn't bring people in at all...

created time in 2 days

issue openedmindee/devrel

Define the newsletter process

Define the newsletter process

Here some notes I took last time

The process will be more fluid next time. Will write a draft on Friday or Monday, will ask people to review on Tuesday, will send a test on Wednesday and will send the newsletter to our customers on Thursday.

created time in 2 days

issue openedmindee/devrel

Use the deep learning behind web API content

Check the "Deep learning behind web API" video Olivier recorded and see what we can do with it. The minimum would be to upload it to YouTube.

created time in 2 days

issue openedmindee/devrel

Add a CTA to "How to use confidence scores in machine learning models" blog post

Add a CTA to "How to use confidence scores in machine learning models" blog post:

  • https://towardsdatascience.com/how-to-use-confidence-scores-in-machine-learning-models-abe9773306fa
  • https://mindee.com/blog/how-use-confidence-scores-ml-models

created time in 2 days

issue openedmindee/devrel

Record a newer/better version of the "Mindee - Build an accurate document parsing API" video

Record a newer/better version of the "Mindee - Build an accurate document parsing API" video

created time in 2 days

issue closedmindee/devrel

Add missing headers images

Can you add featured images for these posts please:

  • https://test.fred.dev/how-to-design-and-deploy-your-first-document-parsing-api/
  • https://test.fred.dev/parsing-receipts-with-mindees-machine-learning-api/
  • https://test.fred.dev/extract-receipt-data-with-mindees-api-using-python/
  • https://test.fred.dev/extending-license-plate-extraction/
  • https://test.fred.dev/parsing-passports-with-mindees-api/

closed time in 2 days

fharper

issue commentmindee/devrel

Add missing headers images

LGTM

fharper

comment created time in 2 days

startedyangyang5214/asciicast2mp4

started time in 2 days

issue openedmindee/doctr

Error when installing requirement on a M1 MacBook

🐛 Bug

When I try to install the requirements to run doctr on my M1 MacBook, I get an ModuleNotFoundError: No module named '_ctypes'. I installed the additional packages required for macOS users.

https://user-images.githubusercontent.com/1170989/133793566-a8b97c2c-55a7-4129-bd95-a4c6535f9798.mp4

To Reproduce

Steps to reproduce the behavior:

  1. pip install -r requirements.txt

Expected behavior

No error when installing the packages.

Environment

DocTR version: N/A TensorFlow version: N/A PyTorch version: N/A (torchvision N/A) OpenCV version: N/A OS: Mac OSX 11.5.2 Python version: 3.7 Is CUDA available (TensorFlow): N/A Is CUDA available (PyTorch): N/A CUDA runtime version: Could not collect GPU models and configuration: Could not collect Nvidia driver version: Could not collect cuDNN version: Could not collect How you installed DocTR (conda, pip, source): I did not, I'm trying to use the code from the repository I cloned

created time in 2 days

startedpytorch/pytorch

started time in 2 days

issue commentmindee/doctr

Cannot run doctr with Docker

You are right, I cannot even install the requirements on my M1 MacBook. I'll create another issue for it as #430 is focussing on arm for low powered devices, while the M1 processor is arm64: AFAIK, it will be two different built, and does not need the same requirements/versions.

fharper

comment created time in 2 days

startedLisiadito/removeInstagramLoginEnforcement

started time in 2 days

issue openedmindee/doctr

Cannot run doctr with Docker

🐛 Bug

I cannot build the doctr Docker image. I get an error when the requirements are being installed: https://asciinema.org/a/i8A8fz50TzdUfZkDcC5DTZW2P

To Reproduce

Steps to reproduce the behaviour:

  1. Build the Docker image
  2. Profit!

Expected behavior

Being able to use doctr with the Dockerfile provided

Environment

OS: Mac OSX 11.5.2 on a M1 chip (it shouldn't change anything since the pip install runs inside the container) Docker: 20.10.8, build 3967b7d

created time in 3 days

startedasciinema/asciicast2gif

started time in 3 days

startedkohler/gifsicle

started time in 3 days

issue commentmindee/devrel

Write new Python SDK version processes

PyPI

First time you need to add the configuration to be able to publish from your computer. In your home directory, create .pypirc file with this content by replacing <TOKEN> with the proper token information for both sites:

[distutils]
index-servers =
    mindee
    mindee-test

[mindee]
repository = https://upload.pypi.org/legacy/
username = __token__
password = <TOKEN>

[mindee-test]
repository = https://test.pypi.org/legacy/
username = __token__
password = <TOKEN>

  • Update setup.py version number
  • Build the source distribution python -m build --sdist
  • Build the wheel python -m build --wheel
  • Upload the package on TestPyPI first python setup.py sdist upload -r minde-test
  • Check that everything is fine
  • Upload the package on PyPi python setup.py sdist upload -r minde
fharper

comment created time in 3 days

issue openedmindee/devrel

Write new Node.js SDK version processes

Write new Node.js SDK version processes. Examples: tag the new version, update the releases notes, push on npm...

created time in 3 days

issue openedmindee/devrel

Write new Python SDK version processes

Write new Python SDK version processes. Examples: tag the new version, update the releases notes, push on PyPI...

created time in 3 days

issue openedmindee/devrel

Update PyPI with SDK v1.2.0

Update PyPI with SDK v1.2.0

created time in 3 days

issue openedmindee/devrel

Create a developer portal

Create a developer portal with https://developers.mindee.com

created time in 3 days

startedmichaellzc/vscode-hadolint

started time in 4 days

issue openedmindee/devrel

Create a newsletter subscribe page

Create a newsletter subscribe page

created time in 4 days

issue openedmindee/devrel

Evaluate our documentation platform

README doesn't seem to be the platform we need. We already had issues with simple things like merging 4 small modifications in one page, and it doesn't handle dynamic documentation like for the API builder...

I suggest we move to Docusaurus. As much as I'm not a fan of Facebook, this is probably the best documentation platform out there, and hosting it ourselves will give us more opportunities to do what we need to do. In addition to having our documentation content on GitHub with a CI/CD pipeline, it will make it easier to:

  • Enforce rules for our documentation;
  • Any technical folks to use their actual Git skills to help with the documentation;
  • Let anyone from our community easily do PR;
  • Since it's OSS and we host it, let us do more than with a SaaS like README.

created time in 5 days

issue openedcode4charity/YouTube-Extension

Video unpause when switching back to the tab even if manually paused

Bug Report:

BUG:
The video start to play again when I switch back to a tab where I manually pause a video (or if it was paused because of the "autopause when switching tab" is activated).

HOW:
Open any video, press play, press pause, switch to another tab, switch back to the tab with the video and the video will unpaused itself.

EXPECTED (/preferred) behavior: The video should not unpaused itself if it was manually paused by the user before switching to another tab. I would also argue that "unpausing when switching back to tab" should be a feature that could be useful for this case instead of this default behaviour, in addition to helping manage even more the "autopause when switching tab" feature.

Setup:

⚬ Browser:
<!-- [e.g. Chromium 83.0.4103.116 / Firefox / Safari / ...] -->

Brave Browser Version 1.29.79 Chromium: 93.0.4577.63 (Official Build) (arm64)

⚬ ImprovedTube Version:

3.263

⚬ Settings:

improvedtube.json.txt

⚬ OS:
<!-- [e.g. Linux Ubuntu 16 / Windows 7 / Mac OSX / iOS ] -->

macOS 11.5.2

created time in 5 days

issue openedmindee/devrel

Answer doctr questions on Bhavesh Bhatt's video

Answer doctr questions on Bhavesh Bhatt's video

created time in 5 days

issue commentmindee/devrel

Review DocTR documentation

As with any feedback I'm giving, keep in mind that I don't have the holy truth so feel free to disagree. I'm usually seeing every little things and since I believe the devil is in the details, I usually share them. I also focus on what could be improved IMNSHO, so don't take anything personal here. Lastly, it's based on the actual knowledge I have of this project, which is basically nothing for now (didn't take the time yet to try it, but fear not my friend, it's coming). If anything is unclear or you want to discuss any points I wrote, please let me know, I'll be more than happy to do so!

General

  • I know doctr is an OSS project, but I think it would make sense to move the doc inside our main documentation. It wouldn't make sense if doctr was not mostly done by us and it wasn't under our own organisation on GitHub. Doing that will ensure the same structure for our users using both, and give some more visibility on our product for doctr users who don't know us.
  • Does our target audience knows already how all this stuff works? If not the doc would need some more "taking the user by the hand" content. Example, in the index you talk about NLU, predictors and other things I have no idea what they are even if I know what OCR is and used some in the past. I may not be the target audience (yet).

https://mindee.github.io/doctr/

  • No need for "Welcome to the documentation of DocTR!"
  • I would change the wording of "Daily maintained" for something like "actively maintained by Mindee" which doesn't promise we push daily (which we nearly do, but we don't), and it's a small promo for us :D
  • Add a Table of Content for the page at the top.
  • No need for changelog & package reference since it's already in the left sidebar.
  • Maybe it's because of my lack of deep knowledge on OCR, but seems like right after the Installation section, we do back into the features of the app. I would move getting started at the end. Even removed it since there is the installation menu to the left.
  • Unless again the target audience knows everything in that Model zoo and supported dataset listing, I may add some more details on the benefits of that.

https://mindee.github.io/doctr/installing.html

  • "This library requires Python 3.6 or higher." should point to Python doc like we do for TensorFlow or PyTorch.
  • "Whichever OS you are running, you will need to install at least TensorFlow or PyTorch. " How can I know which one should I use? Is there specific cases where I should use one or another?
  • "For MacOS users, you can install them as follows" we assume people have Homebrew installed (need to point them in the right direction). We can also proposed a more Python centric way as a alternative with pip.
  • "For Windows users, those dependencies are included in GTK. You can find the latest installer over here." should we point to official GTW Windows installer instead?
  • Add a table of content
  • last code snippet of this page has a formatting problem

Tomorrow I'll check the references.

fg-mindee

comment created time in 5 days