[datasets] Filter currupted and wrong annotated files in ready to use datasets

[felixdittrich92]

Bug description

While testing #933 i have seen some errors (empty crops / non utf-8 strings / and so on)
We need to filter some invalid files/annotations

• ensure all ready to use datasets works fine with eval_detection eval_recognition scripts (TF and PT)
• unify recognition part with recognition dataset / word generator to return string directly instead of {labels: ['string']} update reco datasets and tests #954

detection:

• CORD (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• FUNSD (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• IC03 (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• IC13 (PT/TF) validated by: @felixdittrich92
• IIIT5K (PT/TF) validated by: @felixdittrich92
• IMGUR5K (PT/TF) validated by: @felixdittrich92
• SROIE (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• SVHN (PT/TF) validated by: @felixdittrich92
• SVT (PT/TF) [Fix] SVT dataset: clip box values and add shape and label check #955
• SynthText (PT/TF) validated by: @felixdittrich92

recognition:

• MJSynth (PT/TF) [Fix] MJSynth dataset: filter corrupted or missing images #956
• CORD (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• FUNSD (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• IC03 (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983
• IC13 (PT/TF) validated by: @felixdittrich92
• IIIT5K (PT/TF) validated by: @felixdittrich92
• IMGUR5K (PT/TF) [datasets] Fix recognition parts of SynthText and IMGUR5K #1038
• SROIE (PT/TF) [datasets] update IC / SROIE / FUNSD / CORD #983 (NOTE: 99% contains whitespaces exluding not possible)
• SVHN (PT/TF) [datasets] revert whitespace filtering and fix svhn reco #987
• SVT (PT/TF) [Fix] SVT dataset: clip box values and add shape and label check #955
• SynthText (PT/TF) [datasets] Fix recognition parts of SynthText and IMGUR5K #1038

[felixdittrich92]

@frgfm (only fyi)
If we use polygons there is any memory leak (seems not to come directly from the datasets)
(detection: without use_polygons all datasets works fine)

• profile where it cames from

[frgfm]

@felixdittrich92 oh :/ On a specific dataset? during training?

[felixdittrich92]

@frgfm Unfortunately no it affects all detection datasets used with use_polygons=True.
I only noticed it later because some datasets aren't that big and i haven't had this on track before :/

While testing with eval_detection scripts in both frameworks so it seems to come from any transforms to whole dataset init is fine

[frgfm]

Alright, let's get to the bottom of this 👍

[felixdittrich92]

https://github.com/bloomberg/memray super useful to find the mem leak