Computer Vision – ECCV 2024 Workshops

DARES: Depth Anything in Robotic Endoscopic Surgery with Self-supervised Vector LoRA of the Foundation Model.- LocalMamba: Visual State Space Model with Windowed Selective Scan.- Compositional Text-to-Image Generation with Feedforward Layout Generation.- PackMamba：Efficient Processing of Variable-Length Sequences in Mamba training.- Down-Sampling Inter-Layer Adapter for Parameter and Computation Efficient Ultra-Fine-Grained Image Recognition.- Memory-Efficient Vision Transformers: An Activation-Aware Mixed-Rank Compression Strategy.- LLaMA-NAS: Efficient Neural Architecture Search for Large Language Models.- Improving Hyperparameter Optimization with Checkpointed Model Weights.- MagicDec: Breaking the Latency-Throughput Tradeoff for Long Contexts with Speculative Decoding.- Mixed Non-linear Quantization for Vision Transformers.- CycleBNN: Cyclic Precision Training in Binary Neural Networks.- DailyMAE: Towards Pretraining Masked Autoencoders in One Day.- EPTQ: Enhanced Post-Training Quantization via Hessian-guided Network-wise Optimization.- Generalized SAM: Efficient Fine-Tuning of SAM for Variable Input Image Sizes.- LightAvatar: Efficient Head Avatar as Dynamic Neural Light Field.- Giving each task what it needs - leveraging structured sparsity for tailored multi-task learning.- ERF-NAS: Efficient Receptive Field-based Zero-Shot NAS for Object Detection.- CA3D: Convolutional-Attentional 3D Nets for Efficient Video Activity Recognition on the Edge.- Memory-Optimized Once-For-All network.- Famba-V: Fast Vision Mamba with Cross-Layer Token Fusion.- Latent Distillation for Continual Object Detection at the Edge.- MCUBench: A Benchmark of Tiny Object Detectors on MCUs.- Optimizing Resource Consumption in Diffusion Models through Hallucination Early Detection.