4 ms·
When it weren't for the font it might be anomalies in the image taking or even in the encoder software. You can never really be sure, what exactly the ML is de
by 1718627440 1y ago
When it weren't for the font it might be anomalies in the image taking or even in the encoder software. You can never really be sure, what exactly the ML is detecting.
- osrec 1y agoExactly. A marginally higher image ISO at one location vs a lower ISO at another could potentially have a similar effect, and it would be quite difficult to detect.
- CamperBob2 1y agoYou can give it the same tests the human radiologists take in school. They do take tests, don't they? They don't all score 100% every time, do they?
- 1718627440 1y agoThe point here is that the radiologists has a concept of knowing which light patterns are sensible to draw conclusions from and which not, because the radiologist has a concept of real world 3D objects.
- CamperBob2 1y agoSure. It's just not a valid point. Even if it's valid today, it won't be by next week.
- Avalaxy 1y agoWhy not? That's what Grad-CAM is for right?
- 1718627440 1y agoWhat if the ML takes the conclusion exactly from the right pixels, but the cause is a rasterization issue.