EgoBabyVLM: Benchmarking Cross-Modal Learning from Naturalistic Egocentric Video Data

TL;DR

This paper introduces EgoBabyVLM, a benchmark and challenge for evaluating and advancing vision-language models on naturalistic egocentric video data, highlighting current models' limitations in weakly-aligned, real-world scenarios.

Contribution

It presents a new benchmark suite and evaluation pipeline for models trained on egocentric videos, emphasizing the gap between current models and human-like language grounding in naturalistic settings.

Findings

Current VLMs rely on tightly aligned data and struggle with weakly-aligned egocentric input.

The Machine-DevBench benchmark measures lexical and grammatical competence across frequency bins.

Models trained on naturalistic data underperform compared to those trained on curated datasets.

Abstract

Children acquire language grounding with remarkable robustness from limited visuo-linguistic input in ways that surpass today's best large multimodal models. Recent research suggests current vision-language models (VLMs) trained on curated web data fail to generalize to the sparse, weakly-aligned egocentric streams produced by wearable devices, embodied agents, and infant head-cams -- and no fixed evaluation pipeline exists for measuring progress on this regime. We train VLMs on datasets with varying degrees of semantic alignment between visual and linguistic inputs, including naturalistic infant and adult egocentric videos, and evaluate them with a comprehensive suite spanning multimodal language grounding and unimodal vision and language tasks. At the core of this suite is Machine-DevBench, a corpus-grounded benchmark of lexical and grammatical competence, automatically generated from the model's training vocabulary across logarithmic frequency bins to eliminate the train/eval mismatch and low statistical power of prior developmental benchmarks. Our results show that current VLM paradigms hinge on the tight semantic alignment of curated data and fail to exploit the weakly-aligned signal that dominates naturalistic egocentric input -- the very regime in which humans thrive. To motivate progress, we introduce the EgoBabyVLM Challenge to drive the development of models capable of grounded language learning from the kind of naturalistic data that human infants experience.