Papers › HPTQ: Hardware-Friendly Post Training Quantization

HPTQ: Hardware-Friendly Post Training Quantization

19 Sep 2021arXiv:2109.09113archive 2025-07-28

Hai Victor Habi, Reuven Peretz, Elad Cohen, Lior Dikstein, Oranit Dror, Idit Diamant, Roy H. Jennings, Arnon Netzer

Neural network quantization enables the deployment of models on edge devices. An essential requirement for their hardware efficiency is that the quantizers are hardware-friendly: uniform, symmetric, and with power-of-two thresholds. To the best of our knowledge, current post-training quantization methods do not support all of these constraints simultaneously. In this work, we introduce a hardware-friendly post training quantization (HPTQ) framework, which addresses this problem by synergistically combining several known quantization methods. We perform a large-scale study on four tasks: classification, object detection, semantic segmentation and pose estimation over a wide variety of network architectures. Our extensive experiments show that competitive results can be obtained under hardware-friendly constraints.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

sony/model_optimization officialmentioned in papermentioned on GitHubpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Object DetectionPose EstimationQuantizationSemantic Segmentationobject-detection

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Quantization COCO (Common Objects in Context) SSD ResNet50 V1 FPN 640x640 MAP 34.3 #1 of 1 Archive leaderboard report
Quantization ImageNet Xception W8A8 Activation bits 8 #8 of 27 Archive leaderboard report
Quantization ImageNet Xception W8A8 Top-1 Accuracy (%) 78.972 #8 of 27 Archive leaderboard report
Quantization ImageNet Xception W8A8 Weight bits 8 #8 of 27 Archive leaderboard report
Quantization ImageNet EfficientNet-B0 ReLU W8A8 Activation bits 8 #11 of 27 Archive leaderboard report
Quantization ImageNet EfficientNet-B0 ReLU W8A8 Top-1 Accuracy (%) 77.092 #11 of 27 Archive leaderboard report
Quantization ImageNet EfficientNet-B0 ReLU W8A8 Weight bits 8 #11 of 27 Archive leaderboard report
Quantization ImageNet EfficientNet-B0 W8A8 Activation bits 8 #16 of 27 Archive leaderboard report
Quantization ImageNet EfficientNet-B0 W8A8 Top-1 Accuracy (%) 74.216 #16 of 27 Archive leaderboard report
Quantization ImageNet EfficientNet-B0 W8A8 Weight bits 8 #16 of 27 Archive leaderboard report
Quantization ImageNet DenseNet-121 W8A8 Activation bits 8 #19 of 27 Archive leaderboard report
Quantization ImageNet DenseNet-121 W8A8 Top-1 Accuracy (%) 73.356 #19 of 27 Archive leaderboard report
Quantization ImageNet DenseNet-121 W8A8 Weight bits 8 #19 of 27 Archive leaderboard report
Quantization ImageNet MobileNetV2 W8A8 Activation bits 8 #23 of 27 Archive leaderboard report
Quantization ImageNet MobileNetV2 W8A8 Top-1 Accuracy (%) 71.46 #23 of 27 Archive leaderboard report
Quantization ImageNet MobileNetV2 W8A8 Weight bits 8 #23 of 27 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections