Papers › Grouped Pointwise Convolutions Reduce Parameters in Convolutional Neural Networks
Grouped Pointwise Convolutions Reduce Parameters in Convolutional Neural Networks
Joao Paulo Schwarz Schuler, Santiago Romani, Mohamed Abdel-Nasser, Hatem Rashwan, Domenec Puig
In Deep Convolutional Neural Networks (DCNNs), the parameter count in pointwise convolutions quickly grows due to the multiplication of the filters and input channels from the preceding layer. To handle this growth, we propose a new technique that makes pointwise convolutions parameter-efficient via employing parallel branching, where each branch contains a group of filters and processes a fraction of the input channels. To avoid degrading the learning capability of DCNNs, we propose interleaving the filters' output from separate branches at intermediate layers of successive pointwise convolutions. To demonstrate the efficacy of the proposed technique, we apply it to various state-of-the-art DCNNs, namely EfficientNet, DenseNet-BC L100, MobileNet and MobileNet V3 Large. The performance of these DCNNs with and without the proposed method is compared on CIFAR-10, CIFAR-100, Cropped-PlantDoc and Oxford-IIIT Pet datasets. The experimental results demonstrated that DCNNs with the proposed technique, when trained from scratch, obtained similar test accuracies to the original EfficientNet and MobileNet V3 Large architectures while saving up to 90% of the parameters and 63% of the floating-point computations.
Code
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
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Image Classification | CIFAR-10 | kEffNet-B0 32ch | Percentage correct | 93.75 | #169 of 265 | Archive leaderboard | report |
| Image Classification | CIFAR-10 | kMobileNet V3 Large 16ch | Percentage correct | 92.74 | #179 of 265 | Archive leaderboard | report |
| Image Classification | CIFAR-10 | kDenseNet-BC L100 12ch | Percentage correct | 90.83 | #198 of 265 | Archive leaderboard | report |
| Image Classification | CIFAR-10 | kMobileNet 16ch | Percentage correct | 89.81 | #207 of 265 | Archive leaderboard | report |
| Image Classification | CIFAR-100 | kMobileNet V3 Large 16ch | PARAMS | 0.52M | #171 of 211 | Archive leaderboard | report |
| Image Classification | CIFAR-100 | kMobileNet V3 Large 16ch | Percentage correct | 71.36 | #171 of 211 | Archive leaderboard | report |
| Image Classification | Oxford-IIIT Pet Dataset | kMobileNet V3 Large 16ch | Accuracy | 60.4 | #4 of 5 | Archive leaderboard | report |
| Image Classification | Oxford-IIIT Pet Dataset | kMobileNet V3 Large 16ch | FLOPS | 81.0M | #4 of 5 | Archive leaderboard | report |
| Image Classification | Oxford-IIIT Pet Dataset | kMobileNet V3 Large 16ch | PARAMS | 0.36M | #4 of 5 | Archive leaderboard | report |
| Image Classification | PlantDoc | kMobileNet V3 Large 16ch | PARAMS | 0.43M | #2 of 2 | 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.
Methods
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