charles/monoloco
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

fix mode flags

Browse Source
This commit is contained in:
Lorenzo 2021-04-14 15:39:29 +02:00
parent e21b438b0e
commit 6299859d95
3 changed files with 14 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
README.md
View File

@ -186,7 +186,6 @@ python -m monoloco.run predict docs/frame0032.jpg \
## C) Orientation and Bounding Box dimensions
The network estimates orientation and box dimensions as well. Results are saved in a json file when using the command
`--output_types json`. At the moment, the only visualization including orientation is the social distancing one.
<br />
## Training
@ -195,12 +194,12 @@ We train on the KITTI dataset (MonoLoco/Monoloco++/MonStereo) or the nuScenes da
Results for MonoLoco++ are obtained with:
```
python -m monoloco.run train --joints data/arrays/joints-kitti-201202-1743.json --save --monocular
python -m monoloco.run train --joints data/arrays/joints-kitti-201202-1743.json
```
While for the MonStereo ones just change the input joints and remove the monocular flag:
While for the MonStereo ones just change the input joints and add `--mode stereo`
```
python3 -m monoloco.run train --joints <json file path> --save`
python3 -m monoloco.run train --joints data/arrays/joints-kitti-201202-1022.json --mode stereo
```
If you are interested in the original results of the MonoLoco ICCV article (now improved with MonoLoco++), please refer to the tag v0.4.9 in this repository.
@ -244,12 +243,17 @@ python -m openpifpaf.predict \
--checkpoint=shufflenetv2k30 \
--instance-threshold=0.05 --seed-threshold 0.05 --force-complete-pose
```
Once the step is complete, the below commands transform all the annotations into a single json file that will used for training
Once the step is complete, the below commands transform all the annotations into a single json file that will used for training.
For MonoLoco++:
```
python -m monoloco.run prep --dir_ann <directory that contains annotations>
```
!Add the flag `--monocular` for MonoLoco(++)!
For MonStereo:
```
python -m monoloco.run prep --mode stereo --dir_ann <directory that contains annotations>
```
### Collective Activity Dataset
To evaluate on of the [collective activity dataset](http://vhosts.eecs.umich.edu/vision//activity-dataset.html)

5
monoloco/eval/generate_kitti.py
View File

@ -33,7 +33,8 @@ class GenerateKitti:
# Load Network
assert args.mode in ('mono', 'stereo'), "mode not recognized"
self.net = 'monstereo' if args.mode == 'mono' else 'monoloco_pp'
self.mode = args.mode
self.net = 'monstereo' if args.mode == 'stereo' else 'monoloco_pp'
use_cuda = torch.cuda.is_available()
device = torch.device("cuda" if use_cuda else "cpu")
self.model = Loco(
@ -153,7 +154,7 @@ class GenerateKitti:
print("\nSaved in {} txt {} annotations. Not found {} images".format(cnt_file, cnt_ann, cnt_no_file))
if self.net == 'monstereo':
if self.baselines[self.mode] and self.net == 'monstereo':
print("STEREO:")
for key in self.baselines['stereo']:
print("Annotations corrected using {} baseline: {:.1f}%".format(

2
monoloco/train/trainer.py
View File

@ -85,7 +85,7 @@ class Trainer:
self.mt_loss = MultiTaskLoss(losses_tr, losses_val, self.lambdas, self.tasks)
self.mt_loss.to(self.device)
if not self.mode == 'stereo':
if self.mode == 'stereo':
input_size = 68
output_size = 10
else: