Cs2 Demo Archive: Per-Round Player Video and State Data from CS2 Replays

Sample archives produced by the open-source cs2-demo-to-dataset pipeline, which converts Counter-Strike 2 .dem replay files into structured data. The pipeline outputs per-round, per-player first-person video aligned to tick-level state, input, and event tables. This release by Vasy7777 is intended to demonstrate the pipeline's reproducible archive format rather than serve as a full dataset.

Use Cases

• Training computer vision models for in-game object detection based on aligned first-person video.
• Analyzing player tactics and decision-making based on tick-level state and event tables.
• Developing reinforcement learning agents by simulating environments from recorded game state and input sequences.
• Benchmarking multi-modal alignment techniques using synchronized video and structured game state data.

Strengths

• Data is produced by a documented, open-source pipeline (cs2-demo-to-dataset), suggesting a reproducible format.
• Output includes multiple aligned data modalities: first-person video, player state, inputs, and game events.
• The sample demonstrates a coherent archive structure for complex game replay conversion.

Limitations

• This is explicitly a sample release for pipeline demonstration; the scale and completeness as a dataset are unknown.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license information are unknown, which may limit suitability assessment.

Provenance

Source

Counter-Strike 2 .dem replay files processed via the cs2-demo-to-dataset pipeline.

Collection Method

Conversion of a single game replay file into structured, multi-modal archives.

Freshness

Last updated 2026-06-20 14:00:03; freshness should be verified.