MIMALLOC_PURGE_DELAY=0 only bites on Windows (bundled mimalloc). On a Linux
server PyTorch uses glibc ptmalloc, which keeps a BiRefNet@2048 forward's freed
activations in the arena, so RSS ratchets up across requests and the env var is
a no-op there.
A. bgfilter/memtune.py: release_freed_memory() calls glibc malloc_trim(0) at
runtime (no-op on Windows/musl/other allocators). Wired into service.process
(after each request) and the CLI batch loop (after each image), the two
long-lived paths where RSS accumulates. Single-image CLI exits, so it is left
alone.
B. DEPLOY(_ZH): document preloading jemalloc via LD_PRELOAD + MALLOC_CONF (also
improves CPU throughput) as the production alternative, with MALLOC_ARENA_MAX/
MALLOC_TRIM_THRESHOLD_ as an allocator-free fallback. Under jemalloc/tcmalloc
malloc_trim simply no-ops.
Also fixes the stale off-by-default cross-check heading in DEPLOY.md.
Verified: modules import; release_freed_memory() returns False (no-op) on Windows.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Cross-check + reuse-as-seg are already the config/CLI default; align the HTTP
service with them instead of forcing cross-check off. The earlier off-by-default
was to avoid the extra inference, but reuse-as-seg now makes that forward double
as the seg mask (birefnet backend) -- it replaces the primary seg model rather
than adding to it, so the cost concern is gone.
- app.py: honour the config default (on); BGFILTER_CROSS_CHECK still forces
either way, unset = config.
- service.py preload: warm the cross-check HR-matting model, and skip the now-
redundant primary segmenter when reuse covers segmentation.
- README / DEPLOY(_ZH): document cross-check as default-on, BiRefNet_HR-matting
as required (not optional), BGFILTER_CROSS_CHECK=0 to disable.
Verified: service defaults cross_check on, env forces both ways, preload picks
cross-checker and skips the primary segmenter.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adopts master as the standard for all overlapping work. Master's landed
optimizations supersede server-edition's own fp16 experiment:
- Unified --precision fp32|bf16 knob (bgfilter/precision.py) driving all
three models: ViTMatte weight cast + both BiRefNets via autocast, with a
hardware gate (falls back to fp32 off native-bf16 CPUs) and an
AutocastCPU fp32 shim for torchvision deform_conv2d.
- Query-chunked ViTMatte global attention (bgfilter/attn_chunk.py), exact
and bitwise-identical, caps the N^2 spike (~19 -> ~4 GB at 2048).
- Cross-check HR-matting forward reused as the seg mask (birefnet backend
only), skipping the primary seg model; MIMALLOC_PURGE_DELAY=0.
- inference_mode and the detect_background_color removal converge with
server-edition's earlier equivalents.
Conflict resolution (favoring master, preserving server-only features):
- vitmatte_infer/segmentation: dropped server's device-derived fp16 for
master's precision path, kept resolve_model_source (local weights).
- service.py: cross-check SegmentationSettings now passes precision so the
HTTP service honors bf16 like the CLI's _second_opinion does.
Verified on CPU: default fp32 pipeline loads 2 models (seg-reuse active)
and bf16 path runs (autocast + deform_conv2d shim) — both exit 0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
POST /remove-background gains an optional cross_check form field
(1/0/true/false/yes/no/on/off; empty = service default). The effective
state is reported in the X-BGFilter-Cross-Check response header.
The cross-check model is now never preloaded at startup: the first
request that enables it pays the one-time load (measured ~8 s on top of
the ~50 s cc inference), after which the instance is cached and shared.
Verified with a functional test against PipelineManager: cc-off requests
never load the model, cc=1 lazy-loads then reuses the cached instance,
and omitting the field falls back to the service default.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Brings in the two algorithm optimisations from master:
- Gate chroma suppression by the matte's own confidence (639fdf0)
- Add cross-model veto of hair-gap background residue (adbbb9d)
Server-edition integration on top of the plain merge:
- PipelineManager caches the cross-check model (BiRefNet_HR-matting) and
injects it into MattingPipeline, so HTTP requests never reload it.
- MattingResult carries cross_check_alpha; debug output matches master.
- /healthz reports crossCheck; README lists models/BiRefNet_HR-matting
in the local-weights bundle.
Verified: CLI --no-cross-check baseline and full cross-check run both
pass on samples/TestImage3.png; cross_check_alpha.png written to debug.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
lhk229