Markus Pobitzer

DepthART. [ACMMM 2026] Official code for DepthART: Depth Anything Rethought for Tiny Models

SITH. [CVPR 2026] From Weights to Concepts: Data-Free Interpretability of CLIP via Singular Vector Decomposition

PoInit-of-View. [CVPR-26🎉🎉🎉] This is the official repository for "PoInit-of-View: Poisoning Initialization of Views Transfers Across Multiple 3D Reconstruction Systems".

ram. Python

drawio. draw.io is a JavaScript, client-side editor for general diagramming.

AdvFLYP. [CVPR-26 finding 🎉] This is the official repository for our work "Finetune Like You Pretrain: Boosting Zero-shot Adversarial Robustness in Vision-language Models".

EAGC. (CVPR2026 Highlight) The Devil Is in Gradient Entanglement: Energy-Aware Gradient Coordinator for Robust Generalized Category Discovery (EAGC)

VoRA. [Fully open] [Encoder-free MLLM] Vision as LoRA

proactivebench. Official repository of "ProactiveBench: Benchmarking Proactiveness in Multimodal Large Language Models" (ECCV 2026)

UWM. [WACV 26] Official code for the paper Safe Vision-Language Models via Unsafe Weights Manipulation

mLLMs_merging_4_DMO. Official implementation of the paper "Linear Model Merging Unlocks Simple and Scalable Multimodal Data Mixture Optimization".

MemCoach. [CVPR'26 Highlight] MemCoach: Steering-based MLLM for Actionable Image Memorability Feedback

CIRCLE. [CVPR Findings 2026] Large Multimodal Models as General In-Context Classifiers

maskclu. (AAAI2026, Oral) Masked Clustering Prediction for Unsupervised Point Cloud Pre-training

MLLM-Semantic-Hallucination. 🔥🔥[NeurIPS2025]Exploring and mitigating semantic hallucinations in scene text perception and reasoning

OWDFA-CAL. (AAAI2026) Open-World Deepfake Attribution via Confidence-Aware Asymmetric Learning (CAL)

WebToEpub. A simple Chrome (and Firefox) Extension that converts Web Novels (and other web pages) into an EPUB.

pjax. A projection-based framework for gradient-free and parallel learning

MCLD. [CVPR 2025] Multi-focal Conditioned Latent Diffusion for Person Image Synthesis

CLIP-Test-time-Counterattacks. [CVPR-25🔥] Test-time Counterattacks (TTC) towards adversarial robustness of CLIP

VideoDeepResearch. Python

DEMO. [3DV 2026] Dense Motion Captioning

moviepy. Video editing with Python

BiRefNet. [CAAI AIR'24] Bilateral Reference for High-Resolution Dichotomous Image Segmentation

transformers. 🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

convisbench. Official implementation of "ConViS-Bench: Estimating Video Similarity Through Semantic Concepts", NeurIPS 2025

windows-kernel-driver-pipeline. This master thesis project continuously collects and analyses Microsoft Windows kernel drivers using static and dynamic methods to help security researcher in evaluating and uncovering vulnerabilities.

TEMU-VTOFF. [ICLR 2026] "Inverse Virtual Try-On: Generating Multi-Category Product-Style Images from Clothed Individuals"

Earth-Observation-VLMs. 🔥🔥A Family of Multi-Sensor, Multi-Granularity Vision-Language Models for Earth Observation Understanding

Video-XL. 🔥🔥First-ever hour scale video understanding models

stable-diffusion. A latent text-to-image diffusion model

custom-diffusion. Custom Diffusion: Multi-Concept Customization of Text-to-Image Diffusion (CVPR 2023)

generative-models. Generative Models by Stability AI

fiftyone. Refine high-quality datasets and visual AI models

Pillow. Python Imaging Library (fork)

RePaint. Official PyTorch Code and Models of "RePaint: Inpainting using Denoising Diffusion Probabilistic Models", CVPR 2022

diffusers. 🤗 Diffusers: State-of-the-art diffusion models for image, video, and audio generation in PyTorch.

dphpc. Project for DPHPC

PerceptualSimilarity. LPIPS metric. pip install lpips