An overview of human-machine collaborative image compression methods in literature. MBID, MBHD, SBMD and SBAR respectively represent multi-bitstream independent decoding, multi-bitstream hierarchical decoding, single-bitstream multi-head decoding, and single-bitstream analysis after reconstruction. The ✓ indicates that the method aims to reconstruct and analyze facial images.
| Category | Author | Presented Task | Core Method | Facial Image Specific |
|---|---|---|---|---|
| MBID | [111] | recognition | TFQI-based joint bit allocation | |
| [177] | classification | cross-layer context model + ROI | ||
| [27] | segmentation | semantic feature enhancement | ||
| [24] | detection | slimmable compressive encoder | ||
| [120] | detection segmentation | gate module+knowledge distillation | ||
| MBHD | [9] | face detection | feature map + CNN | ✓ |
| [185] | face recognition | feature & texture representation | ✓ | |
| [131] | facial identity recognition, facial attribute prediction | StyleGAN prior + layer-wise scalable entropy transformer | ✓ | |
| [184] | face verification | feature & texture + residual | ✓ | |
| [80] | facial landmark detection | edge map + GAN | ✓ | |
| [205] | segmentation | GAN+hyperprior model | ✓ | |
| [30] | detection | instance segmentation map + signal feature | ||
| [5] | image search | semantic segmentation map + residual | ||
| [72] | semantic enhancement | semantic segmentation + enhancement | ||
| [190] | classification | task feature+residual | ||
| [104] | classification | residual enhance + GAN | ||
| [165] | detection | ob ject separation + parameter share | ||
| [56] | segmentation, pose estimation | customized group mask + group-independent transform | ||
| [121] | classification | pyramid of multiple subbands | ||
| [54] | face recognition | Canny edge color sketch | ✓ | |
| [202] | detection, segmentation | structural representation+VGG | ||
| [196] | detection | depth-constrained encoder | ||
| [122] | classification, detection, segmentation | hyperprior network + predictor module | ||
| [37] | detection | latent space transform | ||
| [213] | classification, segmentation | reconstruction semantic feature fusion | ||
| [55] | detection, segmentation | structural edges + feature + prior | ||
| [105] | classification | semantics-based ROI mask + generation module | ||
| [38] | detection, segmentation | ask-dependent latent space transform | ||
| [195] | detection | mask multilayer fusion | ||
| [14] | classification | lightweight image encoder+ViT | ||
| SBMD | [123] | classification | general feature extraction + feature-analytic classifier | |
| [26] | classification, detection, segmentation | prompt generator + Transformer | ||
| [176] | classification, segmentation | feature-maps | ||
| SBAR | [132] | face recognition | sketches thumbnails + retrieved guidance | |
| [186] | detection | inverted bottleneck structure encoder | ||
| [62] | detection, segmentation, facial landmark detection | content-adaptive diffusion model | ||
| [59] | image caption, detection | feature distance + importance-weighted pixel distance |
| Category | Author | Presented Task | Core Method | Facial Image Specific |
|---|---|---|---|---|
| MBID | [ | recognition | TFQI-based joint bit allocation | |
| [ | classification | cross-layer context model + ROI | ||
| [ | segmentation | semantic feature enhancement | ||
| [ | detection | slimmable compressive encoder | ||
| [ | detection segmentation | gate module+knowledge distillation | ||
| MBHD | [ | face detection | feature map + CNN | ✓ |
| [ | face recognition | feature & texture representation | ✓ | |
| [ | facial identity recognition, facial attribute prediction | StyleGAN prior + layer-wise scalable entropy transformer | ✓ | |
| [ | face verification | feature & texture + residual | ✓ | |
| [ | facial landmark detection | edge map + GAN | ✓ | |
| [ | segmentation | GAN+hyperprior model | ✓ | |
| [ | detection | instance segmentation map + signal feature | ||
| [ | image search | semantic segmentation map + residual | ||
| [ | semantic enhancement | semantic segmentation + enhancement | ||
| [ | classification | task feature+residual | ||
| [ | classification | residual enhance + GAN | ||
| [ | detection | ob ject separation + parameter share | ||
| [ | segmentation, pose estimation | customized group mask + group-independent transform | ||
| [ | classification | pyramid of multiple subbands | ||
| [ | face recognition | Canny edge color sketch | ✓ | |
| [ | detection, segmentation | structural representation+VGG | ||
| [ | detection | depth-constrained encoder | ||
| [ | classification, detection, segmentation | hyperprior network + predictor module | ||
| [ | detection | latent space transform | ||
| [ | classification, segmentation | reconstruction semantic feature fusion | ||
| [ | detection, segmentation | structural edges + feature + prior | ||
| [ | classification | semantics-based ROI mask + generation module | ||
| [ | detection, segmentation | ask-dependent latent space transform | ||
| [ | detection | mask multilayer fusion | ||
| [ | classification | lightweight image encoder+ViT | ||
| SBMD | [ | classification | general feature extraction + feature-analytic classifier | |
| [ | classification, detection, segmentation | prompt generator + Transformer | ||
| [ | classification, segmentation | feature-maps | ||
| SBAR | [ | face recognition | sketches thumbnails + retrieved guidance | |
| [ | detection | inverted bottleneck structure encoder | ||
| [ | detection, segmentation, facial landmark detection | content-adaptive diffusion model | ||
| [ | image caption, detection | feature distance + importance-weighted pixel distance |
Sharing content requires targeting cookies to be enabled. Please update your cookie preferences to use this feature.