Saved in:
Bibliographic Details
Main Authors: Reddy, N K Sagar, Kasera, Neeraj, Thakur, Avinash
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2403.06895
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910362569277440
author Reddy, N K Sagar
Kasera, Neeraj
Thakur, Avinash
author_facet Reddy, N K Sagar
Kasera, Neeraj
Thakur, Avinash
contents Our research focuses on the analysis and improvement of the Graph-based Relation Inference Transformer (GRIT), which serves as an important benchmark in the field. We conduct a comprehensive ablation study using the PISC-fine dataset, to find and explore improvement in efficiency and performance of GRITv2. Our research has provided a new state-of-the-art relation recognition model on the PISC relation dataset. We introduce several features in the GRIT model and analyse our new benchmarks in two versions: GRITv2-L (large) and GRITv2-S (small). Our proposed GRITv2-L surpasses existing methods on relation recognition and the GRITv2-S is within 2% performance gap of GRITv2-L, which has only 0.0625x the model size and parameters of GRITv2-L. Furthermore, we also address the need for model compression, an area crucial for deploying efficient models on resource-constrained platforms. By applying quantization techniques, we efficiently reduced the GRITv2-S size to 22MB and deployed it on the flagship OnePlus 12 mobile which still surpasses the PISC-fine benchmarks in performance, highlighting the practical viability and improved efficiency of our model on mobile devices.
format Preprint
id arxiv_https___arxiv_org_abs_2403_06895
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle GRITv2: Efficient and Light-weight Social Relation Recognition
Reddy, N K Sagar
Kasera, Neeraj
Thakur, Avinash
Computer Vision and Pattern Recognition
Our research focuses on the analysis and improvement of the Graph-based Relation Inference Transformer (GRIT), which serves as an important benchmark in the field. We conduct a comprehensive ablation study using the PISC-fine dataset, to find and explore improvement in efficiency and performance of GRITv2. Our research has provided a new state-of-the-art relation recognition model on the PISC relation dataset. We introduce several features in the GRIT model and analyse our new benchmarks in two versions: GRITv2-L (large) and GRITv2-S (small). Our proposed GRITv2-L surpasses existing methods on relation recognition and the GRITv2-S is within 2% performance gap of GRITv2-L, which has only 0.0625x the model size and parameters of GRITv2-L. Furthermore, we also address the need for model compression, an area crucial for deploying efficient models on resource-constrained platforms. By applying quantization techniques, we efficiently reduced the GRITv2-S size to 22MB and deployed it on the flagship OnePlus 12 mobile which still surpasses the PISC-fine benchmarks in performance, highlighting the practical viability and improved efficiency of our model on mobile devices.
title GRITv2: Efficient and Light-weight Social Relation Recognition
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2403.06895