Computation Reallocation for Object Detection

Abstract

The allocation of computation resources in the backbone is a crucial issue in object detection. However, classification allocation pattern is usually adopted directly to object detector, which is proved to be sub-optimal. In order to reallocate the engaged computation resources in a more efficient way, we present CR-NAS (Computation Reallocation Neural Architecture Search) that can learn computation reallocation strategies across different feature resolution and spatial position diectly on the target detection dataset. A two-level reallocation space is proposed for both stage and spatial reallocation. A novel hierarchical search procedure is adopted to cope with the complex search space. We apply CR-NAS to multiple backbones and achieve consistent improvements. Our CR-ResNet50 and CR-MobileNetV2 outperforms the baseline by 1.9% and 1.7% COCO AP respectively without any additional computation budget. The models discovered by CR-NAS can be equiped to other powerful detection neck/head and be easily transferred to other dataset, e.g. PASCAL VOC, and other vision tasks, e.g. instance segmentation. Our CR-NAS can be used as a plugin to improve the performance of various networks, which is demanding.

ICLR2020

References in corpus (7)

• Neural Architecture Search with Reinforcement Learning
• SGDR: Stochastic Gradient Descent with Warm Restarts
• DARTS: Differentiable Architecture Search
• Understanding the Effective Receptive Field in Deep Convolutional Neural Networks
• DetNet: A Backbone network for Object Detection
• Single Path One-Shot Neural Architecture Search with Uniform Sampling
• NAS-FCOS: Fast Neural Architecture Search for Object Detection

Cited by in corpus (16)

• A Comprehensive Survey of Neural Architecture Search: Challenges and Solutions
• Sample and Computation Redistribution for Efficient Face Detection
• Blockwisely Supervised Neural Architecture Search with Knowledge Distillation
• PSViT: Better Vision Transformer via Token Pooling and Attention Sharing
• Dilated SpineNet for Semantic Segmentation
• Evolving Search Space for Neural Architecture Search
• TransNAS-Bench-101: Improving Transferability and Generalizability of Cross-Task Neural Architecture Search
• Cheaper Pre-training Lunch: An Efficient Paradigm for Object Detection
• ViPNAS: Efficient Video Pose Estimation via Neural Architecture Search
• Representation Sharing for Fast Object Detector Search and Beyond
• Inception Convolution with Efficient Dilation Search
• Efficient Scale-Permuted Backbone with Learned Resource Distribution
• Real Image Super Resolution Via Heterogeneous Model Ensemble using GP-NAS
• BN-NAS: Neural Architecture Search with Batch Normalization
• Efficient Transfer Learning via Joint Adaptation of Network Architecture and Weight
• NAS-FCOS: Efficient Search for Object Detection Architectures