mmselfsup
mmcv
mmselfsup | mmcv | |
---|---|---|
5 | 4 | |
3,089 | 5,611 | |
0.8% | 1.2% | |
5.3 | 7.7 | |
11 months ago | 12 days ago | |
Python | Python | |
Apache License 2.0 | Apache License 2.0 |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
mmselfsup
-
MMDeploy: Deploy All the Algorithms of OpenMMLab
MMSelfSup: OpenMMLab self-supervised learning toolbox and benchmark.
-
Does anyone know how a loss curve like this can happen? Details in comments
For some reason, the loss goes up shaply right at the start and slowly goes back down. I am self-supervised pretraining an image modeling with DenseCL using mmselfsup (https://github.com/open-mmlab/mmselfsup). This shape happened on the Coco-2017 dataset and my custom dataset. As you can see, it happens consistently for different runs. How could the loss increase so sharply and is it indicative of an issue with the training? The loss peaks before the first epoch is finished. Unfortunately, the library does not support validation.
- Defect Detection using RPI
- [D] State-of-the-Art for Self-Supervised (Pre-)Training of CNN architectures (e.g. ResNet)?
- Rebirth! OpenSelfSup is upgraded to MMSelfSup
mmcv
-
MMDeploy: Deploy All the Algorithms of OpenMMLab
MMCV: OpenMMLab foundational library for computer vision.
- Mmcv - Openmmlab computer vision foundation
-
An elegant and strong PyTorch Trainer
I opened source some works (AAAI 21 SeqNet, ICCV 21 MAED, etc) and earned more than 500 stars. After referring to some popular projects (detectron2, pytorch-image-models, and mmcv), based on my personal development experience, I developed a SIMPLE enough, GENERIC enough, and STRONG enough PyTorch Trainer: core-pytorch-utils, also named CPU. CPU covers most details in the process of training a deep neural network, including:
-
Why do practitioners still use regular tensorflow? [D]
Pretty much any custom layer, loss, ops, etc. For some of the most common ones used for objection detection, see here, examples include rotated iou/nms, deformable convolutions, focal loss variants, sync batch norm, etc.
What are some alternatives?
Unsupervised-Semantic-Segmentation - Unsupervised Semantic Segmentation by Contrasting Object Mask Proposals. [ICCV 2021]
pytorch-image-models - PyTorch image models, scripts, pretrained weights -- ResNet, ResNeXT, EfficientNet, NFNet, Vision Transformer (ViT), MobileNet-V3/V2, RegNet, DPN, CSPNet, Swin Transformer, MaxViT, CoAtNet, ConvNeXt, and more
anomalib - An anomaly detection library comprising state-of-the-art algorithms and features such as experiment management, hyper-parameter optimization, and edge inference.
TensorFlow2.0_Notebooks - Implementation of a series of Neural Network architectures in TensorFow 2.0
calibrated-backprojection-network - PyTorch Implementation of Unsupervised Depth Completion with Calibrated Backprojection Layers (ORAL, ICCV 2021)
pytorch-lightning - Pretrain, finetune and deploy AI models on multiple GPUs, TPUs with zero code changes.
mmagic - OpenMMLab Multimodal Advanced, Generative, and Intelligent Creation Toolbox. Unlock the magic 🪄: Generative-AI (AIGC), easy-to-use APIs, awsome model zoo, diffusion models, for text-to-image generation, image/video restoration/enhancement, etc.
detectron2 - Detectron2 is a platform for object detection, segmentation and other visual recognition tasks.
barlowtwins - Implementation of Barlow Twins paper
mmrotate - OpenMMLab Rotated Object Detection Toolbox and Benchmark
Revisiting-Contrastive-SSL - Revisiting Contrastive Methods for Unsupervised Learning of Visual Representations. [NeurIPS 2021]
aiqc - End-to-end deep learning on your desktop or server.