F2D-SIFPNet: a frequency 2D Slow-I-Fast-P network for faster compressed video action recognition.
Yue MingJiangwan ZhouXia JiaQingfang ZhengLu XiongFan FengNannan HuPublished in: Appl. Intell. (2024)
Keyphrases
- action recognition
- compressed video
- human actions
- human detection
- action classification
- activity recognition
- computer vision
- body parts
- recognizing human actions
- compressed domain
- video streams
- super resolution
- video content
- spatial temporal
- bitstream
- video quality
- wireless sensor networks
- coding scheme
- human activities
- video frames
- data hiding
- human pose
- object recognition
- machine learning