Trending

Bookmarks

Keep Up With Latest Trending Papers. Computer Science, AI and Machine Learning and more.Subscribe

Machine Learning

Computer Vision

Exploiting Real-World Images for Motor Control

Masked Visual Pre-training for Motor Control

We show that self-supervised visual pre-training from real-world images is effective for learning motor control tasks from pixels.

Without relying on labels, state-estimation, or expert demonstrations, we consistently outperform supervised encoders by up to 80% absolute success rate, sometimes even matching the oracle state performance.

To accelerate progress in learning from pixels, we contribute a benchmark suite of hand-designed tasksvarying in movements, scenes, and robots.

Authors

Tete Xiao, Ilija Radosavovic, Trevor Darrell, Jitendra Malik

Related Topics

Reinforcement learning

Visual representations

Neural network controllers

Self-supervised model

Self-supervised visual pre-training

Supervised encoders

Masked modeling of

Read the Paper

Read Paper

Trending

Bookmarks

◐ Recommended Follows

Founder 42papers.com. Also bui...

More

◐ Latest Activity

More

Exploiting Real-World Images for Motor Control - 42Papers

Read Related Papers

Temporal Modeling via Video Transformer and Masked Visual-token Modeling

A generative adversarial network based image transformation policy for visual self-supervision

Data Selection Based Fine-tuning Pipeline for Image Classification

Hierarchical Graph-Convolutional Variational AutoEncoding for Generative Modelling of Human Motion

NÜWA: Visual Synthesis Pre-training for Neural visUal World creAtion

Scaling Up Vision-Language Pre-training for Image Captioning

Deep Reinforced Attention Regression for Partial Sketch Based Image Retrieval

Self-supervised Semi-supervised Learning for Data Labeling and Quality Evaluation

Distribution Auto-Encoder for Action Quality Assessment from Videos

Contrast-reconstruction Representation Learning for Self-supervised Skeleton-based Action Recognition