Research map / Detection and segmentation / CNN features
An Efficient Compressive Convolutional Network for Unified Object Detection and Image Compression
Xichuan Zhou, Lang Xu, Shujun Liu, Yingcheng Lin, Lei Zhang, Cheng Zhuo
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 |
|---|---|---|---|
| Add incoherence regularization to first conv layer (CCN) vs random convolution | CCN-YOLO · VOC image compression | PSNR (dB): 22.21 -> 26.54 | Improved |
| Learned CCN embedding vs data-driven ADAGIO | CCN-YOLO · BSD100 | PSNR (dB): 22.42 -> 26.56 | Improved |
| Use the best original YOLOv2 kernel for embedding vs random convolution | YOLOv2 kernel · image compression | PSNR (dB): 2.1 dB higher | Improved |
| Add incoherence regularization to the best YOLOv2 kernel (ablation) | CCN · image compression | PSNR (dB): 25.3 -> 26.5 | Improved |
| Incoherence regularization during training iterations (ablation) | CCN · VOC training | PSNR (dB): increases by 1.2 dB to over 4.0 dB | Improved |
| Add image compression via CCN to YOLOv2 detection | CCN-YOLO · VOC 2007+2012 | mAP: 78.6 -> 78.1 | No clear effect |
| Add image compression via CCN to YOLOv2 detection | CCN-YOLO · COCO trainval | mAP: 44.0 -> 43.8 | No clear effect |
| Add CCN to YOLOv2 | CCN-YOLO · VOC | recall: 86.8% -> 86.6% | Hurt |
| Add CCN to YOLOv2 | CCN-YOLO · VOC top-5 classes | average accuracy: 93.3% -> 92.8% | Hurt |
| Unified CCN vs separate compressive sensing module plus YOLOv2 | CCN-YOLO · block size 16*16 | inference time (s): 8.81 -> 2.82 | Improved |
| Unified CCN vs Gauss+YOLOv2 at larger block | CCN-YOLO · block size 128*128 | inference time (s): 14.15 -> 2.82 | Improved |
| Backprop-trained CCN vs ADAGIO as training set grows | CCN-YOLO · BSD100 training time | training time (s): 45.782 -> 31.069 | Mixed |
Closest papers by meaning
- Learning Efficient Binarized Object Detectors With Information Compression (TPAMI 2021 · similarity 0.86)
- Sequential Convolution and Runge-Kutta Residual Architecture for Image Compressed Sensing (ECCV 2020 · similarity 0.86)
- Sparse R-CNN: End-to-End Object Detection With Learnable Proposals (CVPR 2021 · similarity 0.86)
- Sparse R-CNN: An End-to-End Framework for Object Detection (TPAMI 2023 · similarity 0.86)
- CPP-Net: Embracing Multi-Scale Feature Fusion into Deep Unfolding CP-PPA Network for Compressive Sensing (CVPR 2024 · similarity 0.84)
- UFC-Net: Unrolling Fixed-point Continuous Network for Deep Compressive Sensing (CVPR 2024 · similarity 0.84)
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.
