菜单
Generate descriptive text prompts from images by combining BLIP captioning with CLIP cosine similarity against curated label banks for medium, movement, and flavor attributes
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Memory-efficient Keras generator that streams sharded CSV files in chunks, renders strokes to images on-the-fly, and yields batches for training on datasets too large for memory
Stack 1D convolutions for local feature extraction before bidirectional LSTMs to classify variable-length stroke sequences into hundreds of doodle categories
Partition a large dataset into N balanced shards using integer key modulo arithmetic for reproducible, class-interleaved splits across CSV files
Normalize raw stroke coordinates to 0-255 range, resample at uniform arc-length spacing, then apply Ramer-Douglas-Peucker simplification
Render stroke sequences to grayscale images with temporal intensity encoding where earlier strokes are brighter and later strokes fade to encode drawing order
Score each face in a video as anomalous by computing L2 distance from the embedding centroid of all faces, then convert to probability via logistic function
| name | cv-clip-interrogator-captioning |
| description | Generate descriptive text prompts from images by combining BLIP captioning with CLIP cosine similarity against curated label banks for medium, movement, and flavor attributes |
When you need to reverse-engineer or describe an image as a text prompt (e.g., for image-to-prompt tasks), CLIP Interrogator combines two models: BLIP generates a base caption, then CLIP matches the image embedding against precomputed text embeddings from curated label banks (mediums, movements, flavors). The top-matching labels are appended to the caption, producing a rich prompt that captures style, medium, and content.
import torch
from clip_interrogator import Config, Interrogator
ci = Interrogator(Config(clip_model_name="ViT-L-14/openai"))
cos = torch.nn.CosineSimilarity(dim=1)
mediums_features = torch.stack([torch.from_numpy(t) for t in ci.mediums.embeds]).to(ci.device)
movements_features = torch.stack([torch.from_numpy(t) for t in ci.movements.embeds]).to(ci.device)
flavors_features = torch.stack([torch.from_numpy(t) for t in ci.flavors.embeds]).to(ci.device)
def interrogate(image):
caption = ci.generate_caption(image)
feat = ci.image_to_features(image)
medium = ci.mediums.labels[cos(feat, mediums_features).topk(1).indices[0]]
movement = ci.movements.labels[cos(feat, movements_features).topk(1).indices[0]]
flavors = ", ".join([ci.flavors.labels[i] for i in cos(feat, flavors_features).topk(3).indices])
return f"{caption}, {medium}, {movement}, {flavors}"
clip_interrogator