CLIP meets GamePhysics: Towards bug identification in gameplay videos using zero-shot transfer learning

TL;DR

This paper introduces a zero-shot video search method using CLIP to identify game bugs in gameplay videos without requiring labeled data, facilitating bug analysis and reproduction from large video repositories.

Contribution

It presents a novel zero-shot approach leveraging CLIP for bug detection in gameplay videos, along with a new dataset of nearly 27,000 videos for evaluation.

Findings

Effective retrieval of relevant gameplay videos using simple and compound queries

Demonstrated capability to detect objects and events related to bugs in videos

Approach does not require labeled training data or metadata

Abstract

Gameplay videos contain rich information about how players interact with the game and how the game responds. Sharing gameplay videos on social media platforms, such as Reddit, has become a common practice for many players. Often, players will share gameplay videos that showcase video game bugs. Such gameplay videos are software artifacts that can be utilized for game testing, as they provide insight for bug analysis. Although large repositories of gameplay videos exist, parsing and mining them in an effective and structured fashion has still remained a big challenge. In this paper, we propose a search method that accepts any English text query as input to retrieve relevant videos from large repositories of gameplay videos. Our approach does not rely on any external information (such as video metadata); it works solely based on the content of the video. By leveraging the zero-shot transfer capabilities of the Contrastive Language-Image Pre-Training (CLIP) model, our approach does not require any data labeling or training. To evaluate our approach, we present the $\texttt{GamePhysics}$ dataset consisting of 26,954 videos from 1,873 games, that were collected from the GamePhysics section on the Reddit website. Our approach shows promising results in our extensive analysis of simple queries, compound queries, and bug queries, indicating that our approach is useful for object and event detection in gameplay videos. An example application of our approach is as a gameplay video search engine to aid in reproducing video game bugs. Please visit the following link for the code and the data: https://asgaardlab.github.io/CLIPxGamePhysics/