Research map / Detection and segmentation / CNN features
Pixel Transposed Convolutional Networks
Hongyang Gao, Hao Yuan, Zhengyang Wang, Shuiwang Ji
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace deconvolution with PixelDCL | U-Net · PASCAL 2012 (from scratch) | mean IOU: 0.415178 -> 0.455972 | Improved |
| replace deconvolution with iPixelDCL | U-Net · PASCAL 2012 (from scratch) | mean IOU: 0.415178 -> 0.448817 | Improved |
| replace deconvolution with PixelDCL | U-Net · MSCOCO 2015 (from scratch) | mean IOU: 0.349769 -> 0.371805 | Improved |
| replace deconvolution with iPixelDCL | U-Net · MSCOCO 2015 (from scratch) | mean IOU: 0.349769 -> 0.360216 | Improved |
| replace deconvolution with iPixelDCL | U-Net · MSCOCO 2015 (from scratch) | pixel accuracy: 0.809327 -> 0.809239 | No clear effect |
| replace deconvolution with iPixelDCL | DeepLab-ResNet · PASCAL 2012 (fine-tuned) | mean IOU: 0.727036 -> 0.738552 | Improved |
| replace deconvolution with PixelDCL | DeepLab-ResNet · PASCAL 2012 (fine-tuned) | mean IOU: 0.727036 -> 0.735585 | Improved |
| use PixelDCL instead of iPixelDCL | U-Net · PASCAL 2012 (from scratch) | mean IOU: 0.448817 -> 0.455972 | Improved |
| replace deconvolution with PixelDCL in VAE decoder | VAE · CelebA generation | checkerboard artifacts | Improved |
| replace deconvolution with PixelDCL | U-Net · PASCAL 2012 timing | training time (10 epochs): 365m26s -> 464m31s | Hurt |
| replace deconvolution with iPixelDCL | U-Net · PASCAL 2012 timing | prediction time: 2m42s -> 4m13s | Hurt |
Closest papers by meaning
- Decoders Matter for Semantic Segmentation: Data-Dependent Decoding Enables Flexible Feature Aggregation (CVPR 2019 · similarity 0.86)
- GENERATING HIGH FIDELITY IMAGES WITH SUBSCALE PIXEL NETWORKS AND MULTIDIMENSIONAL UPSCALING (ICLR 2019 · similarity 0.85)
- EfficientFCN: Holistically-guided Decoding for Semantic Segmentation (ECCV 2020 · similarity 0.84)
- The Devil is in the Decoder: Classification, Regression and GANs (IJCV 2019 · similarity 0.84)
- Generalized Decoding for Pixel, Image, and Language (CVPR 2023 · similarity 0.84)
- DCAN: Dual Channel-wise Alignment Networks for Unsupervised Scene Adaptation (ECCV 2018 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
