Comments (8)
@CaesarPan thanks for your interest in our work. Proposals are generated by off-the-shelf object detectors (e.g., Faster R-CNN) on any arbitrary input video. Bounding box annotations are the ground-truth bounding box coordinates annotated by our in-house annotators. During testing, box annotations are used as grounding evaluation reference.
from grounded-video-description.
@LuoweiZhou Thanks for your fast reply. Just for clarification, during training the box annotations are also used as supervision right?
from grounded-video-description.
Yes, for the supervised methods. The unsupervised baseline has no box supervision during training.
from grounded-video-description.
But for the region classification loss as well as the grounding loss, they also require box supervision. Is that correct?
from grounded-video-description.
Yes, all three proxy tasks: supervised attention, supervised grounding, and region classification.
from grounded-video-description.
Yes, so if we only remove supervised attention for the unsupervised model, I think it still needs box supervision or no?
from grounded-video-description.
No, the unsupervised baseline has none of the three proxy tasks.
from grounded-video-description.
Got it. Thanks very much for your time!
from grounded-video-description.
Related Issues (20)
- Masked Transformer in this repo HOT 3
- In inference-only mode, how to split my own video into several segments to represent each event? HOT 1
- Language evaluation: segments in test_2 and val_2 HOT 4
- rgb_motion_1d features HOT 4
- Error running main.py HOT 3
- Uniform sampling script? HOT 1
- conda environment HOT 2
- How to extract frame-wise feats? HOT 1
- error when running the Starter code with GPU HOT 2
- main.py error HOT 7
- Viewing results HOT 1
- question about use pre-train model on my own video HOT 11
- i can't fine the grd_reference file HOT 1
- main.py HOT 1
- AssertionError HOT 3
- evaluate
- Does process exist to use pre-trained model on my own video?
- About default loss weights
- Import "evaluate" could not be resolved
- Broken dependencies
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from grounded-video-description.