swanlab.Video

Github Source Code

Video(
    data_or_path: str,
    caption: str = None,
) -> None

Parameter	Description
data_or_path	(str) Accepts a video file path. Currently only supports GIF format files.
caption	(str) The label for the video. Used to mark the video when displayed in the experiment dashboard.

Introduction

Convert video data to be recorded by swanlab.log(). Currently only supports GIF format video files.

Format Limitation

Currently swanlab.Video only supports GIF format file paths. Other video formats are not supported yet.

Logging GIF Video Files

Logging a single GIF video:

import swanlab

swanlab.init()
video = swanlab.Video("path/to/video.gif", caption="Training Process Demo")
swanlab.log({"video": video})

Logging multiple GIF videos:

import swanlab

swanlab.init()

examples = []
for i in range(3):
    video = swanlab.Video(f"video_{i}.gif", caption=f"Video Example {i}")
    examples.append(video)

swanlab.log({"examples": examples})

Creating GIF Video Example

Here's a complete example of creating a GIF animation and logging it:

import os.path
import random
from PIL import Image as PILImage
from PIL import ImageDraw
import swanlab

swanlab.init()

# Create a GIF animation
def create_mock_gif(output_path, width=200, height=200, frames=10, duration=100):
    """
    Create a simple mock GIF animation
    
    Parameters:
        output_path: Output GIF file path
        width: Image width (pixels)
        height: Image height (pixels)
        frames: Number of animation frames
        duration: Display time per frame (milliseconds)
    """
    images = []
    
    for i in range(frames):
        # Create a new RGB image
        img = PILImage.new('RGB', (width, height), color=(255, 255, 255))
        draw = ImageDraw.Draw(img)
        
        # Generate random colors
        r = random.randint(0, 255)
        g = random.randint(0, 255)
        b = random.randint(0, 255)
        
        # Draw a random circle on the image
        x = random.randint(0, width)
        y = random.randint(0, height)
        radius = random.randint(10, min(width, height) // 2)
        draw.ellipse([x - radius, y - radius, x + radius, y + radius], fill=(r, g, b))
        
        # Add current frame to the list
        images.append(img)
    
    # Save as GIF animation
    images[0].save(output_path, save_all=True, append_images=images[1:], duration=duration, loop=0)

# Create GIF file
gif_path = "test.gif"
create_mock_gif(gif_path, width=300, height=300, frames=15, duration=200)

# Log to SwanLab
swanlab.log({"video": swanlab.Video(gif_path, caption="This is a test video")})

Use Cases

swanlab.Video is suitable for the following scenarios:

1. Training Process Visualization: Record dynamic changes during model training
2. Experiment Result Display: Show dynamic demonstrations of experiment results
3. Data Visualization: Convert time series data into animated displays
4. Model Behavior Analysis: Record dynamic responses of models under different inputs

Notes

1. File Format: Currently only supports GIF format, does not support other video formats like MP4, AVI, etc.
2. File Size: It's recommended to control GIF file size, as overly large files may affect loading speed
3. Frame Rate Control: When creating GIFs, you can control playback speed through the duration parameter
4. File Path: Ensure the provided file path is correct and the file exists