Multimodal Agent MemorySubmit a paper

Image Memory

64 systems and 19 benchmarks whose primary modality is image.

How they represent memory

How T2 systems change it

Representation × management tier

T1 Static Memory 18 · 28% T2 Dynamic Memory 39 · 61% T3 Self-Evolving Memory 7 · 11% Total
Source- and modality-specific 6 5 – 11
Cross-represented explicit 5 9 7 21
Internalized latent & parametric 1 3 – 4
Hybrid 6 22 – 28

Each system counted once, at its highest tier. Circle area shows the value. Hover to compare with chance; click to list the systems.

What its benchmarks test

Directly evaluated, of 19 benchmarks in the survey's Table 2
  • Cross-modal evidenceCE14
  • Long-range evidenceLE17
  • State changesSC9
  • Conflicting memoriesCF4
  • Unanswerable queriesUQ6
  • Memory-grounded actionMA1
  • Fine-grained visual detailFG7
  • Entity trackingET8
  • Visual state changeVC8

Benchmarks

All 64 systems