MMdetection unable to form final test file

I am trying to use MMdetection 2.0 for MASKD object detection. However, I am facing difficulty in creating the test file.
Here is the code that I have written

from mmdet.datasets import build_dataloader
cfg.data.test.test_mode = True
distributed = False
val_dataset = build_dataset(cfg.data.val)
data_loader = build_dataloader(
    val_dataset,
    samples_per_gpu=1,
    workers_per_gpu=1,
    dist=distributed,
    shuffle=False)
from mmcv.runner import load_checkpoint
from mmcv.parallel import MMDataParallel, MMDistributedDataParallel
from mmdet.apis import single_gpu_test
device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')

model = build_detector(cfg.model, train_cfg=None, test_cfg=cfg.test_cfg)
checkpoint = load_checkpoint(model, WEIGHTS_FILE, map_location='cpu')

model.CLASSES = dataset.CLASSES

model = MMDataParallel(model, device_ids=[0])
outputs = single_gpu_test(model, data_loader, False, None, 0.5)
val_dataset.format_results(outputs)

However I get the following output

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-122-4158962a1af4> in <module>()
----> 1 val_dataset.format_results(outputs)

2 frames
/content/mmdetection/mmdet/datasets/coco.py in format_results(self, results, jsonfile_prefix, **kwargs)
    359         else:
    360             tmp_dir = None
--> 361         result_files = self.results2json(results, jsonfile_prefix)
    362         return result_files, tmp_dir
    363 

/content/mmdetection/mmdet/datasets/coco.py in results2json(self, results, outfile_prefix)
    291         result_files = dict()
    292         if isinstance(results[0], list):
--> 293             json_results = self._det2json(results)
    294             result_files['bbox'] = f'{outfile_prefix}.bbox.json'
    295             result_files['proposal'] = f'{outfile_prefix}.bbox.json'

/content/mmdetection/mmdet/datasets/coco.py in _det2json(self, results)
    228                     data['bbox'] = self.xyxy2xywh(bboxes[i])
    229                     data['score'] = float(bboxes[i][4])
--> 230                     data['category_id'] = self.cat_ids[label]
    231                     json_results.append(data)
    232         return json_results

IndexError: list index out of range

I guess I am unable to get the category_id but I cant find how to fix that.
Please help

View post 0

Team Up for the challenge

Hi
I am a newcomer and I have done some small projects in Deep RL before. If anyone is interested in teaming up DM at testandplayalltime@gmail.com.

View post 1

Train and testSet Size is Different

Hi bhavesh thanks for pointing that out, changes have been made

View post 0

Starter Notebook

Hi thanks for sharing. we more than welcome community contributions

View post 1

Need to download datasets is necessary? or else any other way

hey you can mount your drive and save it there. you can then load from there at your convinience

View post 0

Baseline - DIBRD

Hi thanks for pointing out. we have made the changes

View post 0

Approach to solving

Hi
Currently im going over the model zoo provided by mmdetection and wanted to ask what are some metrics to decide a model. Also what could be some hyper parameters(pre and post) which could be used to improve the score?

View post 0

A way to improve Decision Tree

Hi
Yes secondary score for the challenge is Root Mean Square error.

View post 1