dave_user
207
edits
Changes
no edit summary
==Introduction==
==Building the dataset==
===Defects generation and acquisition===
{| class="wikitable" style="text-align:center;"
|- style="font-weight:bold;"
|+'''Anomalies, generation process and numerosity'''
! Anomaly
! P&P
! Serigraphy
! Manual
! Numerosity
|-
| Missing
| ✓
|
|
| 21
|-
| Manhattan
|
|
| ✓
| 4
|-
| Shift x-axis
| ✓
|
|
| 51
|-
| Shift y-axis
| ✓
|
|
| 58
|-
| Shift&Rotation<br />(shift x+z-axes)
| ✓
|
|
| 57
|-
| Rotation<br />(z-axis)
| ✓<br />
|
|
| 55
|-
| Under soldering
|
| ✓
|
| All
|-
| Over soldering
|
| ✓
|
| All
|}
===Class subdivision and labelling===
===Soldering regions extraction===
==Data augmentation with image synthesis==
===Generative adversarial networks===
===Progressive GAN implementation===
==Results validation==
{| class="wikitable" style="text-align:center;"
|- style="font-weight:bold;"
! Class
! Synth image
! Resolution <br />(pixel)
! Google Colab <br />(min)
! AWS SageMaker <br />(min)
|-
| rowspan="2" | <br />missing
| full
| style="text-align:left;" | 512 × 512
| ~480
| ~410
|-
| upper/lower
| style="text-align:left;" | 256 × 256
| ~435
| ~310
|-
| rowspan="2" | <br />tombstoning
| full
| style="text-align:left;" | 512 × 512
| ~460
| ~390
|-
| upper/lower
| style="text-align:left;" | 256 × 256
| ~420
| ~300
|}
[[File:Missing synthesized images.png|center|thumb|500x500px|Synthesized images for ''missing'' class]]
[[File:Tombstoning synthesized images.png|center|thumb|500x500px|Synthesized images for ''tombstoning'' class]]
{| style="background:transparent; color:black" border="0" align="center" cellpadding="10px" cellspacing="0px" height="550" valign="bottom"
|- align="center"
|+ '''T-SNE algorithm results for ''missing'' class synthesized images'''
|
|[[File:Missing 512x512.png|thumb|350x350px|512 × 512 resolution ''full'' images]]
|
|[[File:Missing 256x256.png|thumb|350x350px|256 × 256 resolution ''full'' images]]
|
|[[File:Missing upper lower 256x256.png|thumb|350x350px|256 × 256 resolution ''upper'' and ''lower'' region images]]
|}
{| style="background:transparent; color:black" border="0" align="center" cellpadding="10px" cellspacing="0px" height="550" valign="bottom"
|- align="center"
|+ '''T-SNE algorithm results for ''tombstoning'' class synthesized images'''
|
|[[File:Tombstoning 512x512.png|thumb|350x350px|512 × 512 resolution ''full'' images]]
|
|[[File:Tombstoning 256x256.png|thumb|350x350px|256 × 256 resolution ''full'' images]]
|
|[[File:Tombstoning upper lower 256x256.png|thumb|350x350px|256 × 256 resolution ''upper'' and ''lower'' region images]]
|}
==Useful links==
*Tero Karras, Timo Aila, Samuli Laine, Jaakko Lehtinen, [https://arxiv.org/abs/1710.10196 ''Progressive Growing of GANs for Improved Quality, Stability, and Variation''], February 2018.
* Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, Aaron Courville, [https://arxiv.org/abs/1704.00028 ''Improved Training of Wasserstein GANs''], December 2017.
*[https://shop.ipc.org/IPC-A-610E-English-D ''IPC-A-610E: Acceptability of Electronic Assemblies''], a standard developed by IPC, April 2010.
