hpi-www-midl-text-localization text-localization-agent issues

hpi-www-midl-text-localization / text-localization-agent

Python-based program for training an agent for text localization using reinforcement learning with ChainerRL

3 stars 3 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Add a script for evaluation trained agents on datasets (#19)

#27 bakoe closed 5 years ago
0
Add instructions on where to download the ResNet files from the the README.md

#26 bakoe closed 5 years ago
0
Add script to create an image sequence of an agent

#25 J-L-O closed 5 years ago
0
Add script that allows creating an image sequence/video of an agent acting on one specific image

#24 bakoe closed 5 years ago
0
Increase obs_size by one to account for the additional value returned…

#23 J-L-O closed 5 years ago
0
Learn directly from the image instead of the feature extractor

#22 J-L-O opened 5 years ago
0
Use LinearDecayEpsilonGreedy (#20)

#21 bakoe closed 5 years ago
1
Use LinearDecayEpsilonGreedy instead of ConstantEpsilonGreedy

#20 bakoe closed 5 years ago
0
Add script for evaluating an agent on a large dataset

#19 bakoe closed 5 years ago
1
Enable logging the IoU to Tensorboard #17

#18 J-L-O closed 5 years ago
0
Implement proper evaluation values/methods

#17 bakoe closed 5 years ago
0
Maybe change reward to also care about "centeredness" on the target

#16 Kolpa closed 5 years ago
2
Evaluation log feedback

#15 J-L-O closed 5 years ago
1
Log evaluation results to TensorBoard (#10)

#14 bakoe closed 5 years ago
1
Small adjustment for ResNet-152 implementation

#13 FMalerba closed 5 years ago
0
Enable training on the COCO-Text dataset

#12 J-L-O closed 5 years ago
0
Make the agent able to run on the COCO dataset

#11 bakoe opened 5 years ago
1
Make TensorBoard display the evaluation run results separately

#10 bakoe closed 5 years ago
0
Make changes necessary to run the training on the chair's servers (#7)

#9 J-L-O closed 5 years ago
0
Allow logging to TensorBoard (#6)

#8 bakoe closed 5 years ago
1
Run the training on one of the GPUs provided

#7 bakoe closed 5 years ago
0
Make the agent produce log files suitable for TensorBoard

#6 bakoe closed 5 years ago
0
Make code more flexible regarding runtime environment

#5 bakoe closed 5 years ago
0
Add information about how to use the environment into README.md

#4 bakoe closed 5 years ago
1
Add basic implementation of the reinforcement learning agent (using ChainerRL)

#3 bakoe closed 5 years ago
0
Implement reset behavior from the base paper

#2 bakoe closed 5 years ago
1
Evaluate the set of "positive" actions returned by the environment's step() function's info return value

#1 bakoe closed 5 years ago
3