SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2303.15014 · CVPR · 2023

Leveraging Hidden Positives for Unsupervised Semantic Segmentation

Jae-Pil Heo, Wonjun Moon, Hyun Seong, Subeen Lee

arXiv · PDF · Open in the Atlas

Code that ran

1 of 3 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 1 has no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2303.15014.json · badge: /api/ran/2303.15014.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
hynnsk/HP community
Papers with Code archive
1 ran · 2 no recorded run
FunctionStatusWhere it lives
ClusterLookup Ran
ran on a synthesized input
hynnsk/HP/model/STEGO.py
code served (permissive licence) · get_code("ffa50103d0b8fac8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, class construct-then-call prompt, model not recorded) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import torch.nn as nn
import torch.nn.functional as F

class ClusterLookup(nn.Module):

    def __init__(self, dim: int, n_classes: int):
        super(ClusterLookup, self).__init__()
        self.n_classes = n_classes
        self.dim = dim
        self.clusters = torch.nn.Parameter(torch.randn(n_classes, dim))

    def reset_parameters(self):
        with torch.no_grad():
            self.clusters.copy_(torch.randn(self.n_classes, self.dim))

    def forward(self, x, alpha, log_probs=False, is_direct=False):
        if is_direct:
            inner_products = x
        else:
            normed_clusters = F.normalize(self.clusters, dim=1)
            normed_features = F.normalize(x, dim=1)
            inner_products = torch.einsum('bchw,nc->bnhw', normed_features, normed_clusters)
        if alpha is None:
            cluster_probs = F.one_hot(torch.argmax(inner_products, dim=1), self.clusters.shape[0]).permute(0, 3, 1, 2).to(torch.float32)
        else:
            cluster_probs = nn.functional.softmax(inner_products * alpha, dim=1)
        cluster_loss = -(cluster_probs * inner_products).sum(1).mean()
        if log_probs:
            return (cluster_loss, nn.functional.log_softmax(inner_products * alpha, dim=1))
        else:
            return (cluster_loss, cluster_probs)


INIT_ARGS = {
    'dim': 4,
    'n_classes': 3,
}

ARGS = {
    'x': torch.randn(2, 4, 5, 6),
    'alpha': 2.0,
    'log_probs': False,
    'is_direct': False,
}

if __name__ == '__main__':
    model = ClusterLookup(**INIT_ARGS)
    loss, probs = model(**ARGS)
    print(f"Loss shape: {loss.shape}")
    print(f"Probs shape: {probs.shape}")
    print(f"Loss value: {loss.item()}")
Setup block drafted by
a language model, class construct-then-call prompt, model not recorded
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: torch and numpy seeded to 0 by the harness immediately before the class was constructed from INIT_ARGS (fixture_check.py)
DinoFeaturizer Not yet run
unverified (harvested, no recorded run)
hynnsk/HP/model/STEGO.py
code served (permissive licence) · get_code("9b53dfa39ff1e25a")
How it ran

no invocation recorded.

STEGOmodel Not yet run
unverified (harvested, no recorded run)
hynnsk/HP/model/STEGO.py
code served (permissive licence) · get_code("f1680b66d62cb44d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 3 samples lifted from this paper's linked repositories, 3 carry a function identity and 0 do not. Of the 3 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 3 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Dramatic demand for manpower to label pixel-level annotations triggered the advent of unsupervised semantic segmentation. Although the recent work employing the vision transformer (ViT) backbone shows exceptional performance, there is still a lack of consideration for task-specific training guidance and local semantic consistency. To tackle these issues, we leverage contrastive learning by excavating hidden positives to learn rich semantic relationships and ensure semantic consistency in local regions. Specifically, we first discover two types of global hidden positives, taskagnostic and task-specific ones for each anchor based on the feature similarities defined by a fixed pre-trained backbone and a segmentation head-in-training, respectively. A gradual increase in the contribution of the latter induces the model to capture task-specific semantic features. In addition, we introduce a gradient propagation strategy to learn semantic consistency between adjacent patches, under the inherent premise that nearby patches are highly likely to possess the same semantics. Specifically, we add the loss propagating to local hidden positives, semantically similar nearby patches, in proportion to the predefined similarity scores. With these training schemes, our proposed method achieves new state-of-the-art (SOTA) results in COCO-stuff, Cityscapes, and Potsdam-3 datasets. Our code is available at: https://github.com/hynnsk/HP.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2303.15014")
get_code_for_paper("2303.15014")
have("2303.15014")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2303.15014.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2303.15014.svg)](https://syntology.ai/paper/2303.15014)

Connect an agent — have() is free.