# High-resolution image processing

Kimi-VL-A3B-Thinking-2506 is an upgraded version of Kimi-VL-A3B-Thinking, with significant improvements in multimodal reasoning, visual perception and understanding, video scene processing, etc. It supports higher-resolution images and can achieve more intelligent thinking while consuming fewer tokens.

A vision model based on the Vision Transformer architecture, trained using SigLIP (Sigmoid Loss for Language-Image Pretraining), suitable for image understanding tasks.

The EVA CLIP model is a vision-language model based on OpenCLIP and timm model weights, supporting tasks such as zero-shot image classification.

The visual encoder component of DFN5B-CLIP, based on ViT-Huge architecture, trained with 378x378 resolution images for CLIP model

Vision model based on SigLIP image encoder with global average pooling, part of the PaliGemma2 model

This model is a fine-tuned version of openai/clip-vit-large-patch14-336, primarily used for image-text matching tasks.

Idefics2 is an open multimodal model capable of accepting arbitrary sequences of images and text as input and generating text output. The model can answer questions about images, describe visual content, create stories based on multiple images, or function purely as a language model.

InternViT-6B-448px-V1-5 is a vision foundation model fine-tuned based on InternViT-6B-448px-V1-2, featuring strong robustness, OCR capabilities, and high-resolution processing.

Idefics2 is an open-source multimodal model developed by Hugging Face, capable of processing image and text inputs to generate text outputs, excelling in OCR, document understanding, and visual reasoning.

ChatTruth-7B is a multilingual vision-language model optimized based on the Qwen-VL architecture, enhanced with large-resolution image processing capabilities and incorporating a restoration module to reduce computational overhead

A vision Transformer (ViT)-based image feature model pre-trained using self-supervised DINOv2 method on the LVD-142M dataset

A pre-trained model based on Google's Vision Transformer (ViT) architecture, fine-tuned on the EuroSAT dataset, suitable for remote sensing image classification tasks.

A SegFormer semantic segmentation model fine-tuned on the CityScapes dataset at 1024x1024 resolution, featuring a hierarchical Transformer encoder and a lightweight all-MLP decoder head architecture.