Research map / Detection and segmentation / Tracking
"SLAck: Semantic, Location, and Appearance Aware Open-Vocabulary Tracking"
Siyuan Li, Lei Ke, Yung-Hsu Yang, Luigi Piccinelli, Mattia Segù, Martin Danelljan, Luc Van Gool
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add semantic cue to location-only association (ablation) | Lck · TAO open-vocab val novel | AssocA: 28.3 -> 35.4 | Improved |
| add semantic cue to appearance-only association (ablation) | Ack · TAO open-vocab val novel | AssocA: 32.7 -> 35.1 | Improved |
| add semantic cue to location+appearance tracker (ablation) | LAck · TAO open-vocab val novel | AssocA: 36.4 -> 37.8 | Improved |
| detection-aware training instead of training on sparse ground truth (ablation) | SLAck · TAO open-vocab val novel | AssocA: 24.1 -> 37.8 | Improved |
| hard class grouping of semantic cues instead of no semantics (ablation) | TETer-T · TAO TETA | AssocA: 35.2 -> 30.6 | Hurt |
| soft grouping with BERT embeddings (ablation) | TETer-T · TAO TETA | AssocA: 35.2 -> 31.2 | Hurt |
| early semantic-aware matching instead of CEM soft grouping (ablation) | TETer-T · TAO TETA | AssocA: 36.7 -> 38.0 | Improved |
| use semantic cue alone instead of appearance alone in STOG (ablation) | STOG tracker · TAO open-vocab val novel | AssocA: -4.4 AssocA | Hurt |
| add spatial and temporal object graph (STOG) to all three cues (ablation) | SLAck · TAO open-vocab val novel | AssocA: 34.1 -> 37.8 | Improved |
| add STOG to semantic-only tracker (ablation) | Sck · TAO open-vocab val novel | AssocA: 28.7 -> 28.4 | No clear effect |
| add dynamic score thresholding at inference (ablation) | SLAck · TAO open-vocab val novel | LocA: 50.5 -> 54.3 | Improved |
| use addition instead of concatenation for cue fusion (ablation) | SLAck · TAO open-vocab val novel | AssocA: 35.8 -> 37.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- OVTrack — OVTrack: Open-Vocabulary Multiple Object Tracking (CVPR 2023) · better on 6 of 6 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OVTR: End-to-End Open-Vocabulary Multiple Object Tracking with Transformer (ICLR 2025 · similarity 0.92)
- OVT-B: A New Large-Scale Benchmark for Open-Vocabulary Multi-Object Tracking (NeurIPS 2024 · similarity 0.90)
- Foundation Model Driven Appearance Extraction for Robust Multiple Object Tracking (AAAI 2025 · similarity 0.88)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.87)
- Towards Generalizable Multi-Object Tracking (CVPR 2024 · similarity 0.87)
- Focusing on Tracks for Online Multi-Object Tracking (CVPR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
