Research map / Vision-language and multimodal / Human and embodied modeling
GluonCV and GluonNLP: Deep Learning in Computer Vision and Natural Language Processing
Jian Guo, He He, Tong He, Leonard Lausen, Mu Li, Haibin Lin, Xingjian Shi, Chenguang Wang, Junyuan Xie, Sheng Zha, Aston Zhang, Hang Zhang and 4 more
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use GluonCV training refinements (data augmentation, optimization) instead of Caffe reference | ResNet-50 · ImageNet | top-1 accuracy: 75.3 -> 79.2 | Improved |
| use GluonCV training refinements instead of Caffe reference | ResNet-101 · ImageNet | top-1 accuracy: 76.4 -> 80.5 | Improved |
| use GluonCV implementation instead of TensorFlow reference | MobileNet 1.0 · ImageNet | top-1 accuracy: 70.9 -> 73.3 | Improved |
| use GluonCV implementation instead of Detectron | Faster R-CNN · COCO detection | mAP: 39.6 -> 40.1 | Improved |
| use GluonCV implementation instead of Detectron | Mask R-CNN · COCO instance segmentation | mask AP: 32.8 -> 33.1 | Improved |
| use GluonCV implementation instead of mmaction | I3D ResNet-50 · Kinetics400 | top-1 accuracy: 72.9 -> 74.0 | Improved |
| use GluonNLP implementation instead of reference | textCNN · TREC | accuracy: 92.2 -> 92.8 | Improved |
| use GluonNLP implementation instead of Google BERT | BERT-base · MRPC | accuracy: 86.7 -> 88.7 | Improved |
| use GluonNLP implementation instead of Google BERT | BERT-base · MNLI-m | accuracy: 84.4 -> 84.6 | Improved |
| use GluonNLP implementation instead of Google BERT | BERT-base · SST-2 | accuracy: 92.7 -> 93.0 | Improved |
| use GluonNLP implementation instead of Google BERT | BERT-base · SQuAD 1.1 | F1/EM: 88.5/80.8 -> 88.5/81.0 | No clear effect |
| use GluonNLP implementation instead of Google BERT | BERT-large · SQuAD 1.1 | F1/EM: 90.9/84.1 -> 91.0/84.1 | No clear effect |
Closest papers by meaning
- KerasCV and KerasNLP: Multi-framework Models (JMLR 2024 · similarity 0.85)
- Overcoming the Pitfalls of Vision-Language Model Finetuning for OOD Generalization (ICLR 2024 · similarity 0.83)
- Matryoshka Query Transformer for Large Vision-Language Models (NeurIPS 2024 · similarity 0.82)
- CogVLM: Visual Expert for Pretrained Language Models (NeurIPS 2024 · similarity 0.82)
- ECoFLaP: Efficient Coarse-to-Fine Layer-Wise Pruning for Vision-Language Models (ICLR 2024 · similarity 0.82)
- Vision‑Language‑Vision Auto‑Encoder: Scalable Knowledge Distillation from Diffusion Models (NeurIPS 2025 · similarity 0.82)
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.
