Deterministic VGG16 activation motifs beat edge detection; try constraining to B&W during max activation
moultano · x · 2026-10-13
@srconstantin shows a purely deterministic visualization pipeline: extract motifs from VGG16 activations, upscale, add fine detail and texture back, then threshold to black and white. The result is far more legible than plain Laplacian-of-Gaussians edge detection, with no image-gen models involved.
@moultano suggests a refinement: constrain the image to be black and white while finding the max activations, rather than post-processing — harder to do, but likely to yield cooler results.
Related event: Developer Generates Black-and-White Patterns from VGG16 Activation Maps(2 posts)→
More from Research
- Embodied First Exam benchmarks coding agents on 1,090 tasks across 44 robots — ManlingLi_ · 2026-10-23
- Anima Anandkumar on neural operators: AI weather models beat supercomputers — AnimaAnandkumar · 2026-10-23
- Scaling GLM-5.3-Flash training to 1M token context, 64x over Megatron baseline — AndrewLBeam · 2026-10-23
- Experimental boolean-logic LM: microsecond forward pass on one CPU thread, but painfully slow training — cephaloform · 2026-10-23
- AdaDex turns seconds of human hand motion into generalizable dexterous robot skills — mangahomanga · 2026-10-23
- ReAPO paper: conditioning retries on failures beats GRPO in RLVR reasoning training — jasondeanlee · 2026-10-23