X-CLIP Base Model (patch32-16-frames)
| Property | Value |
|---|---|
| Parameter Count | 197M |
| License | MIT |
| Paper | View Paper |
| Top-1 Accuracy | 81.1% |
| Framework | PyTorch |
What is xclip-base-patch32-16-frames?
X-CLIP is a sophisticated extension of the CLIP architecture designed specifically for video-language understanding. This base model variant processes videos using 16 frames at 224x224 resolution with 32-pixel patches. It was developed by Microsoft and trained on the Kinetics-400 dataset, achieving impressive accuracy metrics in video classification tasks.
Implementation Details
The model employs a transformer-based architecture that processes video and text inputs in a contrastive learning framework. It handles video data by analyzing 16 temporal frames, making it efficient for video understanding tasks while maintaining high accuracy.
- Uses 32x32 pixel patches for video frame processing
- Processes 16 frames per video sequence
- Implements contrastive learning between video and text pairs
- Trained on Kinetics-400 dataset
Core Capabilities
- Video classification with 81.1% top-1 and 95.5% top-5 accuracy
- Zero-shot and few-shot video classification
- Video-text retrieval
- Feature extraction for downstream tasks
Frequently Asked Questions
Q: What makes this model unique?
This model uniquely extends CLIP's capabilities to video understanding while maintaining a relatively compact architecture (197M parameters). It achieves high accuracy while processing multiple frames, making it practical for real-world applications.
Q: What are the recommended use cases?
The model excels in video classification tasks, video-text retrieval, and can be used for zero-shot learning scenarios. It's particularly suitable for applications requiring understanding of video content in relation to textual descriptions.
