|
--- |
|
library_name: transformers |
|
tags: [] |
|
--- |
|
|
|
## Original result |
|
``` |
|
IoU metric: bbox |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.000 |
|
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.000 |
|
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.000 |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000 |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.000 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.000 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.003 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.008 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.008 |
|
``` |
|
|
|
## After training result |
|
``` |
|
IoU metric: bbox |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.012 |
|
Average Precision (AP) @[ IoU=0.50 | area= all | maxDets=100 ] = 0.023 |
|
Average Precision (AP) @[ IoU=0.75 | area= all | maxDets=100 ] = 0.011 |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000 |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 |
|
Average Precision (AP) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.012 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 1 ] = 0.056 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets= 10 ] = 0.118 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= all | maxDets=100 ] = 0.145 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= small | maxDets=100 ] = -1.000 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area=medium | maxDets=100 ] = 0.000 |
|
Average Recall (AR) @[ IoU=0.50:0.95 | area= large | maxDets=100 ] = 0.146 |
|
``` |
|
|
|
## Config |
|
- dataset: NIH |
|
- original model: hustvl/yolos-tiny |
|
- lr: 1e-05 |
|
- dropout_rate: 0.1 |
|
- weight_decay: 0.05 |
|
- max_epochs: 30 |
|
- train samples: 885 |
|
|
|
## Logging |
|
### Training process |
|
``` |
|
{'validation_loss': tensor(6.7207, device='cuda:0'), 'validation_loss_ce': tensor(2.1866, device='cuda:0'), 'validation_loss_bbox': tensor(0.5249, device='cuda:0'), 'validation_loss_giou': tensor(0.9547, device='cuda:0'), 'validation_cardinality_error': tensor(98.5312, device='cuda:0')} |
|
{'training_loss': tensor(3.7287, device='cuda:0'), 'train_loss_ce': tensor(1.3098, device='cuda:0'), 'train_loss_bbox': tensor(0.2119, device='cuda:0'), 'train_loss_giou': tensor(0.6797, device='cuda:0'), 'train_cardinality_error': tensor(0.8000, device='cuda:0'), 'validation_loss': tensor(3.4332, device='cuda:0'), 'validation_loss_ce': tensor(1.2399, device='cuda:0'), 'validation_loss_bbox': tensor(0.2065, device='cuda:0'), 'validation_loss_giou': tensor(0.5804, device='cuda:0'), 'validation_cardinality_error': tensor(1.0909, device='cuda:0')} |
|
{'training_loss': tensor(2.8569, device='cuda:0'), 'train_loss_ce': tensor(0.5845, device='cuda:0'), 'train_loss_bbox': tensor(0.2106, device='cuda:0'), 'train_loss_giou': tensor(0.6097, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.4866, device='cuda:0'), 'validation_loss_ce': tensor(0.5491, device='cuda:0'), 'validation_loss_bbox': tensor(0.1759, device='cuda:0'), 'validation_loss_giou': tensor(0.5290, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(2.2491, device='cuda:0'), 'train_loss_ce': tensor(0.5182, device='cuda:0'), 'train_loss_bbox': tensor(0.1662, device='cuda:0'), 'train_loss_giou': tensor(0.4500, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.3079, device='cuda:0'), 'validation_loss_ce': tensor(0.4791, device='cuda:0'), 'validation_loss_bbox': tensor(0.1619, device='cuda:0'), 'validation_loss_giou': tensor(0.5096, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.9377, device='cuda:0'), 'train_loss_ce': tensor(0.5210, device='cuda:0'), 'train_loss_bbox': tensor(0.1273, device='cuda:0'), 'train_loss_giou': tensor(0.3902, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.2293, device='cuda:0'), 'validation_loss_ce': tensor(0.4691, device='cuda:0'), 'validation_loss_bbox': tensor(0.1544, device='cuda:0'), 'validation_loss_giou': tensor(0.4940, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(2.1924, device='cuda:0'), 'train_loss_ce': tensor(0.5255, device='cuda:0'), 'train_loss_bbox': tensor(0.1414, device='cuda:0'), 'train_loss_giou': tensor(0.4800, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.1471, device='cuda:0'), 'validation_loss_ce': tensor(0.4648, device='cuda:0'), 'validation_loss_bbox': tensor(0.1457, device='cuda:0'), 'validation_loss_giou': tensor(0.4770, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(2.4833, device='cuda:0'), 'train_loss_ce': tensor(0.4674, device='cuda:0'), 'train_loss_bbox': tensor(0.2011, device='cuda:0'), 'train_loss_giou': tensor(0.5053, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.1557, device='cuda:0'), 'validation_loss_ce': tensor(0.4630, device='cuda:0'), 'validation_loss_bbox': tensor(0.1464, device='cuda:0'), 'validation_loss_giou': tensor(0.4804, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.9442, device='cuda:0'), 'train_loss_ce': tensor(0.4237, device='cuda:0'), 'train_loss_bbox': tensor(0.1272, device='cuda:0'), 'train_loss_giou': tensor(0.4424, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.1246, device='cuda:0'), 'validation_loss_ce': tensor(0.4547, device='cuda:0'), 'validation_loss_bbox': tensor(0.1406, device='cuda:0'), 'validation_loss_giou': tensor(0.4833, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.8449, device='cuda:0'), 'train_loss_ce': tensor(0.5204, device='cuda:0'), 'train_loss_bbox': tensor(0.1064, device='cuda:0'), 'train_loss_giou': tensor(0.3963, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0914, device='cuda:0'), 'validation_loss_ce': tensor(0.4524, device='cuda:0'), 'validation_loss_bbox': tensor(0.1409, device='cuda:0'), 'validation_loss_giou': tensor(0.4673, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(2.2881, device='cuda:0'), 'train_loss_ce': tensor(0.4765, device='cuda:0'), 'train_loss_bbox': tensor(0.1549, device='cuda:0'), 'train_loss_giou': tensor(0.5186, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.1367, device='cuda:0'), 'validation_loss_ce': tensor(0.4505, device='cuda:0'), 'validation_loss_bbox': tensor(0.1434, device='cuda:0'), 'validation_loss_giou': tensor(0.4846, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.5250, device='cuda:0'), 'train_loss_ce': tensor(0.4858, device='cuda:0'), 'train_loss_bbox': tensor(0.0681, device='cuda:0'), 'train_loss_giou': tensor(0.3494, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0727, device='cuda:0'), 'validation_loss_ce': tensor(0.4480, device='cuda:0'), 'validation_loss_bbox': tensor(0.1342, device='cuda:0'), 'validation_loss_giou': tensor(0.4769, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.5382, device='cuda:0'), 'train_loss_ce': tensor(0.3929, device='cuda:0'), 'train_loss_bbox': tensor(0.1066, device='cuda:0'), 'train_loss_giou': tensor(0.3061, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9921, device='cuda:0'), 'validation_loss_ce': tensor(0.4464, device='cuda:0'), 'validation_loss_bbox': tensor(0.1298, device='cuda:0'), 'validation_loss_giou': tensor(0.4483, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.8638, device='cuda:0'), 'train_loss_ce': tensor(0.4090, device='cuda:0'), 'train_loss_bbox': tensor(0.1044, device='cuda:0'), 'train_loss_giou': tensor(0.4665, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0358, device='cuda:0'), 'validation_loss_ce': tensor(0.4468, device='cuda:0'), 'validation_loss_bbox': tensor(0.1343, device='cuda:0'), 'validation_loss_giou': tensor(0.4588, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.6616, device='cuda:0'), 'train_loss_ce': tensor(0.4866, device='cuda:0'), 'train_loss_bbox': tensor(0.0970, device='cuda:0'), 'train_loss_giou': tensor(0.3450, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0047, device='cuda:0'), 'validation_loss_ce': tensor(0.4417, device='cuda:0'), 'validation_loss_bbox': tensor(0.1302, device='cuda:0'), 'validation_loss_giou': tensor(0.4559, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.4619, device='cuda:0'), 'train_loss_ce': tensor(0.4675, device='cuda:0'), 'train_loss_bbox': tensor(0.1028, device='cuda:0'), 'train_loss_giou': tensor(0.2401, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0066, device='cuda:0'), 'validation_loss_ce': tensor(0.4412, device='cuda:0'), 'validation_loss_bbox': tensor(0.1329, device='cuda:0'), 'validation_loss_giou': tensor(0.4505, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.5520, device='cuda:0'), 'train_loss_ce': tensor(0.4766, device='cuda:0'), 'train_loss_bbox': tensor(0.0950, device='cuda:0'), 'train_loss_giou': tensor(0.3002, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9805, device='cuda:0'), 'validation_loss_ce': tensor(0.4422, device='cuda:0'), 'validation_loss_bbox': tensor(0.1306, device='cuda:0'), 'validation_loss_giou': tensor(0.4428, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(2.0331, device='cuda:0'), 'train_loss_ce': tensor(0.5165, device='cuda:0'), 'train_loss_bbox': tensor(0.1336, device='cuda:0'), 'train_loss_giou': tensor(0.4242, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9927, device='cuda:0'), 'validation_loss_ce': tensor(0.4389, device='cuda:0'), 'validation_loss_bbox': tensor(0.1282, device='cuda:0'), 'validation_loss_giou': tensor(0.4563, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.4602, device='cuda:0'), 'train_loss_ce': tensor(0.2981, device='cuda:0'), 'train_loss_bbox': tensor(0.1021, device='cuda:0'), 'train_loss_giou': tensor(0.3257, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9925, device='cuda:0'), 'validation_loss_ce': tensor(0.4381, device='cuda:0'), 'validation_loss_bbox': tensor(0.1300, device='cuda:0'), 'validation_loss_giou': tensor(0.4522, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.8886, device='cuda:0'), 'train_loss_ce': tensor(0.4785, device='cuda:0'), 'train_loss_bbox': tensor(0.1082, device='cuda:0'), 'train_loss_giou': tensor(0.4345, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9744, device='cuda:0'), 'validation_loss_ce': tensor(0.4333, device='cuda:0'), 'validation_loss_bbox': tensor(0.1284, device='cuda:0'), 'validation_loss_giou': tensor(0.4496, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.8621, device='cuda:0'), 'train_loss_ce': tensor(0.5109, device='cuda:0'), 'train_loss_bbox': tensor(0.0748, device='cuda:0'), 'train_loss_giou': tensor(0.4886, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9745, device='cuda:0'), 'validation_loss_ce': tensor(0.4335, device='cuda:0'), 'validation_loss_bbox': tensor(0.1278, device='cuda:0'), 'validation_loss_giou': tensor(0.4509, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.0607, device='cuda:0'), 'train_loss_ce': tensor(0.4695, device='cuda:0'), 'train_loss_bbox': tensor(0.0499, device='cuda:0'), 'train_loss_giou': tensor(0.1708, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9522, device='cuda:0'), 'validation_loss_ce': tensor(0.4317, device='cuda:0'), 'validation_loss_bbox': tensor(0.1251, device='cuda:0'), 'validation_loss_giou': tensor(0.4474, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.7454, device='cuda:0'), 'train_loss_ce': tensor(0.4005, device='cuda:0'), 'train_loss_bbox': tensor(0.0675, device='cuda:0'), 'train_loss_giou': tensor(0.5036, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9562, device='cuda:0'), 'validation_loss_ce': tensor(0.4312, device='cuda:0'), 'validation_loss_bbox': tensor(0.1280, device='cuda:0'), 'validation_loss_giou': tensor(0.4424, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.1943, device='cuda:0'), 'train_loss_ce': tensor(0.4380, device='cuda:0'), 'train_loss_bbox': tensor(0.0761, device='cuda:0'), 'train_loss_giou': tensor(0.1880, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9790, device='cuda:0'), 'validation_loss_ce': tensor(0.4290, device='cuda:0'), 'validation_loss_bbox': tensor(0.1262, device='cuda:0'), 'validation_loss_giou': tensor(0.4595, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.3419, device='cuda:0'), 'train_loss_ce': tensor(0.3899, device='cuda:0'), 'train_loss_bbox': tensor(0.0794, device='cuda:0'), 'train_loss_giou': tensor(0.2774, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9489, device='cuda:0'), 'validation_loss_ce': tensor(0.4249, device='cuda:0'), 'validation_loss_bbox': tensor(0.1241, device='cuda:0'), 'validation_loss_giou': tensor(0.4519, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.9394, device='cuda:0'), 'train_loss_ce': tensor(0.3998, device='cuda:0'), 'train_loss_bbox': tensor(0.1388, device='cuda:0'), 'train_loss_giou': tensor(0.4227, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0161, device='cuda:0'), 'validation_loss_ce': tensor(0.4221, device='cuda:0'), 'validation_loss_bbox': tensor(0.1321, device='cuda:0'), 'validation_loss_giou': tensor(0.4667, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.3021, device='cuda:0'), 'train_loss_ce': tensor(0.3494, device='cuda:0'), 'train_loss_bbox': tensor(0.0630, device='cuda:0'), 'train_loss_giou': tensor(0.3188, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0121, device='cuda:0'), 'validation_loss_ce': tensor(0.4226, device='cuda:0'), 'validation_loss_bbox': tensor(0.1326, device='cuda:0'), 'validation_loss_giou': tensor(0.4634, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.4117, device='cuda:0'), 'train_loss_ce': tensor(0.4587, device='cuda:0'), 'train_loss_bbox': tensor(0.0724, device='cuda:0'), 'train_loss_giou': tensor(0.2954, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9528, device='cuda:0'), 'validation_loss_ce': tensor(0.4217, device='cuda:0'), 'validation_loss_bbox': tensor(0.1242, device='cuda:0'), 'validation_loss_giou': tensor(0.4550, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.4619, device='cuda:0'), 'train_loss_ce': tensor(0.3838, device='cuda:0'), 'train_loss_bbox': tensor(0.0605, device='cuda:0'), 'train_loss_giou': tensor(0.3878, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(2.0346, device='cuda:0'), 'validation_loss_ce': tensor(0.4194, device='cuda:0'), 'validation_loss_bbox': tensor(0.1323, device='cuda:0'), 'validation_loss_giou': tensor(0.4769, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.2125, device='cuda:0'), 'train_loss_ce': tensor(0.3469, device='cuda:0'), 'train_loss_bbox': tensor(0.0725, device='cuda:0'), 'train_loss_giou': tensor(0.2517, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9540, device='cuda:0'), 'validation_loss_ce': tensor(0.4169, device='cuda:0'), 'validation_loss_bbox': tensor(0.1248, device='cuda:0'), 'validation_loss_giou': tensor(0.4566, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.1738, device='cuda:0'), 'train_loss_ce': tensor(0.4249, device='cuda:0'), 'train_loss_bbox': tensor(0.0353, device='cuda:0'), 'train_loss_giou': tensor(0.2863, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9534, device='cuda:0'), 'validation_loss_ce': tensor(0.4152, device='cuda:0'), 'validation_loss_bbox': tensor(0.1264, device='cuda:0'), 'validation_loss_giou': tensor(0.4531, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
{'training_loss': tensor(1.2029, device='cuda:0'), 'train_loss_ce': tensor(0.4904, device='cuda:0'), 'train_loss_bbox': tensor(0.0577, device='cuda:0'), 'train_loss_giou': tensor(0.2119, device='cuda:0'), 'train_cardinality_error': tensor(1., device='cuda:0'), 'validation_loss': tensor(1.9811, device='cuda:0'), 'validation_loss_ce': tensor(0.4117, device='cuda:0'), 'validation_loss_bbox': tensor(0.1291, device='cuda:0'), 'validation_loss_giou': tensor(0.4620, device='cuda:0'), 'validation_cardinality_error': tensor(1., device='cuda:0')} |
|
``` |
|
|
|
## Examples |
|
{'size': tensor([512, 512]), 'image_id': tensor([1]), 'class_labels': tensor([4]), 'boxes': tensor([[0.2622, 0.5729, 0.0847, 0.0773]]), 'area': tensor([1717.9431]), 'iscrowd': tensor([0]), 'orig_size': tensor([1024, 1024])} |
|
|
|
 |
|
|
