AI Paper Atlas

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

AAAI 2019 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 8 citations (OpenAlex)

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 changedWhereMeasuredEffect
Add incoherence regularization to first conv layer (CCN) vs random convolutionCCN-YOLO · VOC image compressionPSNR (dB): 22.21 -> 26.54Improved
Learned CCN embedding vs data-driven ADAGIOCCN-YOLO · BSD100PSNR (dB): 22.42 -> 26.56Improved
Use the best original YOLOv2 kernel for embedding vs random convolutionYOLOv2 kernel · image compressionPSNR (dB): 2.1 dB higherImproved
Add incoherence regularization to the best YOLOv2 kernel (ablation)CCN · image compressionPSNR (dB): 25.3 -> 26.5Improved
Incoherence regularization during training iterations (ablation)CCN · VOC trainingPSNR (dB): increases by 1.2 dB to over 4.0 dBImproved
Add image compression via CCN to YOLOv2 detectionCCN-YOLO · VOC 2007+2012mAP: 78.6 -> 78.1No clear effect
Add image compression via CCN to YOLOv2 detectionCCN-YOLO · COCO trainvalmAP: 44.0 -> 43.8No clear effect
Add CCN to YOLOv2CCN-YOLO · VOCrecall: 86.8% -> 86.6%Hurt
Add CCN to YOLOv2CCN-YOLO · VOC top-5 classesaverage accuracy: 93.3% -> 92.8%Hurt
Unified CCN vs separate compressive sensing module plus YOLOv2CCN-YOLO · block size 16*16inference time (s): 8.81 -> 2.82Improved
Unified CCN vs Gauss+YOLOv2 at larger blockCCN-YOLO · block size 128*128inference time (s): 14.15 -> 2.82Improved
Backprop-trained CCN vs ADAGIO as training set growsCCN-YOLO · BSD100 training timetraining time (s): 45.782 -> 31.069Mixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy