kevanme commited on
Commit
5427bec
ยท
verified ยท
1 Parent(s): be6fd22

End of training

Browse files
Files changed (1) hide show
  1. README.md +28 -71
README.md CHANGED
@@ -3,95 +3,52 @@ library_name: transformers
3
  license: apache-2.0
4
  base_model: hustvl/yolos-tiny
5
  tags:
6
- - object-detection
7
- - transformers
8
- - vision
9
- - pytorch
10
- - raccoon
11
- - yolos
12
- - fine-tuning
13
- - huggingface
14
  model-index:
15
  - name: practica_2
16
  results: []
17
  ---
18
 
19
- # practica_2 โ€“ YOLOS Tiny fine-tuned on Raccoon Dataset ๐Ÿฆ
 
20
 
21
- This model is a fine-tuned version of [`hustvl/yolos-tiny`](https://huggingface.co/hustvl/yolos-tiny) on the [Raccoon Dataset](https://github.com/datitran/raccoon_dataset), converted to COCO format. It detects **raccoons** in images using a transformer-based object detection architecture.
22
 
23
- ## ๐Ÿง  Model description
24
 
25
- YOLOS ("You Only Look One-level Series") is a pure Transformer-based object detector. This particular model uses the **Tiny** variant of YOLOS as the base, making it lightweight and efficient for quick inference on small datasets or low-resource environments.
26
 
27
- This version has been fine-tuned to detect a single class: **raccoon**.
28
 
29
- ## ๐Ÿ“Œ Intended uses & limitations
30
 
31
- ### Use cases
32
- - Wildlife monitoring (specifically raccoons)
33
- - Educational/demo applications for transformer-based object detection
34
- - Transfer learning starter for similar single-class detection tasks
35
 
36
- ### Limitations
37
- - Trained only to detect raccoons โ€” not suitable for general-purpose detection.
38
- - May underperform on complex or cluttered scenes due to dataset size.
39
- - Limited generalization beyond the training distribution.
40
 
41
- ## ๐Ÿ“‚ Training and evaluation data
42
 
43
- - **Dataset**: [Raccoon Dataset by Dat Tran](https://github.com/datitran/raccoon_dataset)
44
- - **Format**: Converted from Pascal VOC to COCO
45
- - **Size**: ~200 annotated images
46
- - **Split**: 80% training, 20% test
47
 
48
- ## โš™๏ธ Training procedure
49
 
50
- The model was trained using the Hugging Face `Trainer` API with the following settings:
 
 
 
 
 
 
 
 
51
 
52
- ### ๐Ÿงพ Hyperparameters
53
 
54
- - **Base model**: `hustvl/yolos-tiny`
55
- - **Epochs**: 100
56
- - **Train batch size**: 8
57
- - **Learning rate**: 1e-5
58
- - **Weight decay**: 1e-4
59
- - **Mixed precision**: Native AMP (`fp16=True`)
60
- - **Scheduler**: Linear
61
- - **Optimizer**: AdamW (betas=(0.9, 0.999), epsilon=1e-8)
62
 
63
- ### ๐Ÿ–ผ๏ธ Data augmentation
64
 
65
- Applied using Albumentations:
66
- - Resize (480x480)
67
- - Horizontal flip
68
- - Random brightness and contrast
69
 
70
- ### ๐Ÿงช Evaluation
71
-
72
- Evaluation was performed on the 20% test split, but metrics were not included in this version of the model card. You can run custom evaluation using the `Trainer.evaluate()` method.
73
-
74
- ## ๐Ÿ—‚๏ธ Classes
75
-
76
- | ID | Class |
77
- |----|----------|
78
- | 1 | raccoon |
79
-
80
- ## ๐Ÿ“ฆ Framework versions
81
-
82
- - `transformers`: 4.52.2
83
- - `pytorch`: 2.6.0+cu124
84
- - `datasets`: 2.14.4
85
- - `tokenizers`: 0.21.1
86
-
87
- ## โœ๏ธ Citation
88
-
89
- If you use this model, please consider citing the original YOLOS paper:
90
-
91
- ```bibtex
92
- @inproceedings{fang2021you,
93
- title={You Only Look One-level Feature},
94
- author={Fang, Wanli and Yang, Xiaolin and Wang, Qiang},
95
- booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
96
- year={2021}
97
- }
 
3
  license: apache-2.0
4
  base_model: hustvl/yolos-tiny
5
  tags:
6
+ - generated_from_trainer
 
 
 
 
 
 
 
7
  model-index:
8
  - name: practica_2
9
  results: []
10
  ---
11
 
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
 
15
+ # practica_2
16
 
17
+ This model is a fine-tuned version of [hustvl/yolos-tiny](https://huggingface.co/hustvl/yolos-tiny) on the None dataset.
18
 
19
+ ## Model description
20
 
21
+ More information needed
22
 
23
+ ## Intended uses & limitations
24
 
25
+ More information needed
 
 
 
26
 
27
+ ## Training and evaluation data
 
 
 
28
 
29
+ More information needed
30
 
31
+ ## Training procedure
 
 
 
32
 
33
+ ### Training hyperparameters
34
 
35
+ The following hyperparameters were used during training:
36
+ - learning_rate: 1e-05
37
+ - train_batch_size: 8
38
+ - eval_batch_size: 8
39
+ - seed: 42
40
+ - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
41
+ - lr_scheduler_type: linear
42
+ - num_epochs: 100
43
+ - mixed_precision_training: Native AMP
44
 
45
+ ### Training results
46
 
 
 
 
 
 
 
 
 
47
 
 
48
 
49
+ ### Framework versions
 
 
 
50
 
51
+ - Transformers 4.52.2
52
+ - Pytorch 2.6.0+cu124
53
+ - Datasets 2.14.4
54
+ - Tokenizers 0.21.1