akanametov
Follow
Stars
You Only Search Once: On Lightweight Differentiable Architecture Search for Resource-Constrained Embedded Platforms
Efficient vision foundation models for high-resolution generation and perception.
Train ImageNet *fast* in 500 lines of code with FFCV
Retinaface get 80.99% in widerface hard val using mobilenet0.25.
Retinaface get 80.99% in widerface hard val using mobilenet0.25.
Implementation of popular deep learning networks with TensorRT network definition API
Code for "Accelerating Transformer Pre-training with 2:4 Sparsity"
🚀 | Ultra-lightweight 1MB face detection & landmarks models for mobile and edge devices. > ONNX > WebCam
Efficient 2:4 sparse training algorithms and implementations
SCUT HEAD is a large-scale head detection dataset, including 4405 images labeld with 111251 heads.
Enhanced RFW dataset with added gender and age annotations for comprehensive demographic analysis in face recognition research.
Pytorch implementation of our paper accepted by NeurIPS 2022 -- Learning Best Combination for Efficient N:M Sparsity
PyTorch implementation of "Single-Path NAS: Device-Aware Efficient ConvNet Design"
🎓Automatically Update circult-eda-mlsys-tinyml Papers Daily using Github Actions (Update Every 8th hours)
Exploiting Layer-wise N:M Sparsity for Vision Transformer Acceleration
Fast and Practical Neural Architecture Search (ICCV2019)
[CVPR 2023] DepGraph: Towards Any Structural Pruning
[NeurIPS 2023] Structural Pruning for Diffusion Models
Deepak Ghimire, Kilho Lee, and Seong-heum Kim, Loss-aware automatic selection of structured pruning criteria for deep neural network acceleration, Image and Vision Computing, vol. 136, p. 104745, 2…
Awasome Papers and Resources in Deep Neural Network Pruning with Source Code.
OTOv1-v3, NeurIPS, ICLR, TMLR, DNN Training, Compression, Structured Pruning, Erasing Operators, CNN, Diffusion, LLM
Implementations of different neural network pruning techniques