Temporal Action Detection with Structured Segment Networks Supplementary Materials

نویسندگان

  • Yue Zhao
  • Yuanjun Xiong
  • Limin Wang
  • Zhirong Wu
  • Xiaoou Tang
  • Dahua Lin
  • Hong Kong
چکیده

3. Per-Class Detection Performance Although we obtain superior overall detection performance, it may also be of interest for audience to see the per-class performance. Due to space limit in the text, we present the per-class average of AP values using SSN on ActivityNet v1.2 validation set in Table 1. The average AP values are measured by varying the IOU thresholds from 0.5 to 0.95 in the step of 0.05. For comparison, detection results produced by SSN with proposals generated from a sliding window (486 proposals per video, AR = 71%) and TAG (100 proposals per video, AR = 67%) method are listed in parallel, showing that TAG-SSN achieves a higher AP on most of the classes. The results are also visualized in Fig. 3.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Temporal Segment Networks for Action Recognition in Videos

Deep convolutional networks have achieved great success for image recognition. However, for action recognition in videos, their advantage over traditional methods is not so evident. We present a general and flexible video-level framework for learning action models in videos. This method, called temporal segment network (TSN), aims to model long-range temporal structures with a new segment-based...

متن کامل

Supplementary Material for “Subcategory-aware Convolutional Neural Networks for Object Proposals and Detection”

We present additional qualitative examples obtained from our detection framework on the KITTI detection benchmark [2] and the PASCAL3D+ dataset [8] in this supplementary material. For car in KITTI and the 12 rigid categories in PASCAL3D+, we use 3D Voxel Patterns (3DVPs) [7] as subcategories in our region proposal network and our detection network. For pedestrian and cyclist in KITTI, we cluste...

متن کامل

Crop Land Change Monitoring Based on Deep Learning Algorithm Using Multi-temporal Hyperspectral Images

Change detection is done with the purpose of analyzing two or more images of a region that has been obtained at different times which is Generally one of the most important applications of satellite imagery is urban development, environmental inspection, agricultural monitoring, hazard assessment, and natural disaster. The purpose of using deep learning algorithms, in particular, convolutional ...

متن کامل

Temporal Segment Networks: Towards Good Practices for Deep Action Recognition

Deep convolutional networks have achieved great success for visual recognition in still images. However, for action recognition in videos, the advantage over traditional methods is not so evident. This paper aims to discover the principles to design effective ConvNet architectures for action recognition in videos and learn these models given limited training samples. Our first contribution is t...

متن کامل

Improve Accurate Pose Alignment and Action Localization by Dense Pose Estimation

In this work we explore the use of shape-based representations as an auxiliary source of supervision for pose estimation. We show that shape-based representations can act as a source of ‘privileged information’ that complements and extends the pure landmark-level annotations. We explore 2D shape-based supervision signals, such as Support Vector Shape. Our experiments show that shape-based super...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2017