mlsanityData to 14 Oct 2026

Lightweight Delivery Detection on Doorbell Cameras

arXiv 2305.07812 ↗First posted 13 May 2023cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Despite recent advances in video-based action recognition and robust spatio-temporal modeling, most of the proposed approaches rely on the abundance of computational resources to afford running huge and computation-intensive convolutional or transformer-based neural networks to obtain satisfactory results. This limits the deployment of such models on edge devices with limited power and computing resources. In this work we investigate an important smart home application, video based delivery detection, and present a simple and lightweight pipeline for this task that can run on resource-constrained doorbell cameras. Our method relies on motion cues to generate a set of coarse activity proposals followed by their classification with a mobile-friendly 3DCNN network. To train we design a novel semi-supervised attention module that helps the network to learn robust spatio-temporal features and adopt an evidence-based optimization objective that allows for quantifying the uncertainty of predictions made by the network. Experimental results on our curated delivery dataset shows the significant effectiveness of our pipeline and highlights the benefits of our training phase novelties to achieve free and considerable inference-time performance gains.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 38 of its 56 references that are in the map. First posted on arXiv 13 May 2023.

Builds on

38 of 56 references.

References · 38

  1. 2017Attention Is All You Needcited by 60,361
  2. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  4. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,139
  5. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,442
  6. 2017On Calibration of Modern Neural Networkscited by 4,960
  7. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
  8. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,845
  9. 2017Quo Vadis, Action Recognition? A New Model and the Kinetics Datasetcited by 3,270
  10. 2015SSD: Single Shot MultiBox Detectorcited by 3,130
  11. 2017The Kinetics Human Action Video Datasetcited by 1,909
  12. 2018SlowFast Networks for Video Recognitioncited by 1,867
26 more
  1. 2021ViViT: A Video Vision Transformercited by 1,620
  2. 2017ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devicescited by 1,200
  3. 2017A Closer Look at Spatiotemporal Convolutions for Action Recognitioncited by 1,143
  4. 2018ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Designcited by 1,136
  5. 2021Video Swin Transformercited by 1,091
  6. 2014Two-Stream Convolutional Networks for Action Recognition in Videoscited by 1,068
  7. 2016SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model sizecited by 1,060
  8. 2019EfficientDet: Scalable and Efficient Object Detectioncited by 1,020
  9. 2017Simple Online and Realtime Tracking with a Deep Association Metriccited by 910
  10. 2021Multiscale Vision Transformerscited by 897
  11. 2016Simple Online and Realtime Trackingcited by 868
  12. 2018Evidential Deep Learning to Quantify Classification Uncertaintycited by 732
  13. 2021MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformercited by 642
  14. 2018Real-world Anomaly Detection in Surveillance Videoscited by 577
  15. 2017Rethinking Spatiotemporal Feature Learning: Speed-Accuracy Trade-offs in Video Classificationcited by 521
  16. 2019Bag of Tricks and A Strong Baseline for Deep Person Re-identificationcited by 430
  17. 2018Two at Once: Enhancing Learning and Generalization Capacities via IBN-Netcited by 364
  18. 2018Video Action Transformer Networkcited by 316
  19. 2020FastReID: A Pytorch Toolbox for General Instance Re-identificationcited by 166
  20. 2020Improving model calibration with accuracy versus uncertainty optimizationcited by 123
  21. 2021Evidential Deep Learning for Open Set Action Recognitioncited by 96
  22. 2019Resource Efficient 3D Convolutional Neural Networkscited by 50
  23. 2022Transformer-based Models to Deal with Heterogeneous Environments in Human Activity Recognitioncited by 12
  24. 2018A Proposal-Based Solution to Spatio-Temporal Action Detection in Untrimmed Videoscited by 7
  25. 2021Evaluating Transformers for Lightweight Action Recognitioncited by 4
  26. 2019Assisted Excitation of Activations: A Learning Technique to Improve Object Detectorscited by 2

Built on by

Not cited yet.