EBC-JEPA per-event LED denoising: carrier "k groups (equal count)"

Frozen V-JEPA 2.1 tokens on LED, tokenised as k groups (equal count) (each cell's events cut in time order into 8 groups of equal count, re-laid by tubelet), scored per event: each event looks up the token at its (time bin / 2, patch) and is judged with its own E15 neighbourhood patch. Routes in this repo:

  • V12 AC: token + E15 patch CNN, no history (v12_ac/).
  • Integrated: History Reader (kv) conditions the token on earlier chunks, a TemporalViT reads the patch, and Pixel-AC fuses the per-event scores into one decision per pixel and chunk (integrated/<run>/; result.json carries the Pixel-AC metrics with the AC-only metrics under ac). Run-name suffixes: _nokv = reader bypassed, _lb = scalar Local branch on, _vit_base = ViT-B/384 tokens.

All units: 6,000 training chunks, 4 epochs, official 713 chunks at a fixed threshold of 0.5, no selection on test.

unit route backbone kv LB SR NR DA
integrated/equal_count_time History Reader + TemporalViT AC + Pixel-AC ViT-L on off 0.8439 0.8995 0.8717
integrated/equal_count_time_lb TemporalViT AC only (no Pixel-AC) ViT-L on on 0.8415 0.8918 0.8666
integrated/equal_count_time_nokv_lb TemporalViT AC only (no Pixel-AC) ViT-L off on 0.8245 0.9035 0.8640
integrated/equal_count_time_nokv TemporalViT AC only (no Pixel-AC) ViT-L off off 0.8268 0.8972 0.8620
v12_ac V12 AC: event's V-JEPA token + E15 patch CNN, no history ViT-L off off 0.8795 0.8438 0.8617

Reference on the same protocol: the collaborator's ViT-L histogram integrated run with Pixel-AC, DA .8714. Code: EBC-JEPA-share (denoising-local-run) and its share-contrib-vitb-downstream worktree.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Collection including h3lloworld/ebc-jepa-led-equal-count

Evaluation results

  • Denoising accuracy (SR + NR) / 2, threshold 0.5 on LED official test (66 scenes, 713 chunks)
    self-reported
    0.872
  • Signal retention on LED official test (66 scenes, 713 chunks)
    self-reported
    0.844
  • Noise removal on LED official test (66 scenes, 713 chunks)
    self-reported
    0.899