Results

101 Action
Classes Model Active

Classifier: Plain 3D CNN
Weights: ucf101_3dcnn_best.pth

System Alerts

Clear

WebAPI Action Classification module is fully operational
Just Now
PyTorch CUDA execution backend detected
10 Mins Ago
Spatiotemporal Grad-CAM heatmap overlays compiled
2 Hours Ago

English
Spanish
French

Kazi HossainUSW Final Year

[email protected]

Start Here

User Tools

Support

Lovely Unified Mindful Artificial Thought

Generate new Thoughts with Computer Vision and Generative AI

Human Action Recognition

Automatically identify, classify, and spatio-temporally map action sequences in video streams utilizing our deep-learned 3D CNN model. Track neural activation patterns overlayed via Grad-CAM overlays.
Generative AI (Upcoming...)

Simulate deep human dialogue, extract semantic insights, and perform multimodal question answering using conversational large language models specialized on computer vision research domains.

Lovely Unified Mindful Artificial Thought
Dedicated to "Habiba Lumat" - My Guiding Light, My Eternal Love.

13,320 Videos / 101 Classes

279,683 Neural Parameters