Visual Embedding Reduction and Space Exploration — Clustering-guided Insights for Training Data Enhancement in V-rDu
-
Updated
Nov 18, 2025 - Python
Visual Embedding Reduction and Space Exploration — Clustering-guided Insights for Training Data Enhancement in V-rDu
基于 LAVILA 预训练模型,实现第一视角视频和文本描述的特征提取以及相似度计算。
Human Things explores whether human similarity judgments can improve machine vision embeddings by fine-tuning ResNet-50 models on THINGS object images and human odd-one-out similarity data.
internet bullshit avoidance system
AI-powered personal stylist that extracts garment-level representations from real images using DeepFashion2 annotations and CLIP embeddings, forming the foundation for outfit compatibility and trend-aware fashion recommendation.
💠 Ambiente local e aberto, com ferramentas para observar, experimentar, inspecionar e documentar representações visuais de modelos. 💠 Primeira bancada: 🔹 Bancada Visual Qwen3.5-9B: ferramenta local para preparar imagens e sequências e explorar representações da torre visual do Qwen3.5-9B, com os pesos congelados.
To associate your repository with the visual-embeddings topic, visit your repo's landing page and select "manage topics."