EDSSA: An Encoder-Decoder Semantic Segmentation Networks Accelerator on OpenCL-Based FPGA Platform.
Hongzhi HuangYakun WuMengqi YuXuesong ShiFei QiaoLi LuoQi WeiXinjun LiuPublished in: Sensors (2020)
Keyphrases
- semantic segmentation
- field programmable gate array
- street scenes
- video codec
- superpixels
- label transfer
- distributed video coding
- conditional random fields
- scene classification
- rate distortion
- weakly supervised
- video coding
- hardware implementation
- object classes
- object categories
- motion estimation
- object recognition
- object class
- parallel computing
- biologically inspired
- object classification
- bit rate
- multiscale
- image processing
- natural images
- higher order