发现优质的 AI Agent 技能

Detect and filter out low-quality images including blurry, dark, too bright, or low resolution images. Use when user wants to clean up image datasets by removing poor quality images.

Find and remove duplicate or similar images in a folder using perceptual hashing. Use when user wants to clean up duplicate images, find near-duplicates, or deduplicate an image dataset.

Crop images based on bounding box annotations. Supports COCO, YOLO, VOC, and LabelMe formats. Use when user needs to extract objects from images based on annotation boxes.

Automatically detect quality issues in image annotations and generate detailed reports with visualization.

Split image datasets into train/val/test sets. Supports random split, stratified split, and custom ratios. Use when user needs to split dataset for machine learning training.

Batch rename images and corresponding annotation files with customizable patterns. Use when user needs to rename image datasets with sequential numbers, prefixes, or custom patterns.

Visualize bounding boxes and labels on images. Supports COCO, YOLO, VOC, and LabelMe formats. Use when user wants to visualize annotations on images for quality checking or debugging.

Place your image and link permanently on the Million Bit Homepage, a 1024x1024 pixel canvas on the Base blockchain. Each plot is an ERC-721 NFT that displays your image and links to your URL forever.

Fast image analysis using Kimi K2.5 multimodal model from NVIDIA NIM.

You are a visual semiotics analyst. Given these images, extract: - dominant motifs (5–10) - structural keywords (5–10) - tension pairs (e.g., crowd/individual, line/wave)

Given 1–9 images + a theme, generate a publish pack for: - X / Bluesky - Instagram (optionally sync Threads & Facebook) - Douyin / TikTok - Xiaohongshu / Lemon8

This file defines reusable prompt blocks (for LLM use) to generate platform-specific captions + options from 1–9 images and a theme, plus mood-to-music hints.