Notebook ready for first review #6
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:40:58Z Before this there should be a markdown section assuming that people have installed fiftyone and covered the very basics of the application - which probably should point to my basic quickstart.
An alternative is you create a notebook that covers how to install fiftyone and such but that seems like a repetition of the stuff I wrote
Also cover what you expect as their experience level, how long should this take,... You can see some of that here
thesteve0 commented on 2025-04-01T20:47:12Z Oh wait - the diff didn't show all the stuff above - it was collapsed. Sorry about that thesteve0 commented on 2025-04-01T21:04:32Z ignore everything above except for my suggestion for linkning to the install doc or something |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:40:59Z Please add an intro section before this section explaining (and linking) to the fiftyone Dataset. You should cover the importance of Dataset, how it does not contain the actual image but instead has a reference to it.
In the current paragraph: Link to the Fiftyone dataset Zoo page and link to the quickstart dataset in the documentation. There is no permament link for that right now so please just put a TODO in the URL or something so we know to come back and fix it. |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:00Z Line #5. # When we specify persistent = True, we make sure that changes to the dataset are persisted between multiple FiftyOne sessions.
|
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:00Z I think dataset.info might be more informative here. It would also give you a structure to start explaining the components of a dataset.
I think it is important to discuss a bit more about ephemeral versus persistent and when you might want to use each. This is something that catches a lot of users and so we want to make it clear right from the beginning.
Remember to try and add links |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:01Z Have a markdown paragraph here explaining what samples and why they should care. Remember, use links.
Before you get to fields make sure to explain there is information stored at the dataset level and then information at the sample level. Explain that it points to a media file in the filepath attribute. ID, created_at, last_modified_at are all autogenerated and refer to the sample record in the db, not the image |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:02Z There are three types of fields:
In fiftyone fields are typed and their type is immutable
Again links to docs |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:02Z Before you get into labels do a section on tags, show how to add and delete a tag. Link to tag docs |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:03Z Note, both are detections but they have different attributes. Because they are Detection type, fiftyone has methods that will work out of the box with these types - "it knows what to do with them"
Please list all the label type or at least point to the page with all the types |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:04Z This is good to show but I think need to add a sentence or two to explain why this will be useful. |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:05Z Mention that because we set the value equal to True this field is a boolean type and then show the code to examine it's type |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:06Z Move this back up to the dataset section |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:06Z I would suggesting starting with views. That would allow you to show that you can use the view as a filter to create a new clone with just the subset of data that you want to use |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:07Z This is going to need more explanation. But I think you can save that explanation for the filtering lesson. I would skip adding this here altogether - it's too complicated and their brain is probably already close to full |
|
View / edit / reply to this conversation on ReviewNB thesteve0 commented on 2025-04-01T20:41:08Z Move this to a separate notebook on filtering, views, subsetting.... |
|
Oh wait - the diff didn't show all the stuff above - it was collapsed. Sorry about that View entire conversation on ReviewNB |
|
ignore everything above except for my suggestion for linkning to the install doc or something View entire conversation on ReviewNB |
No description provided.