# How to decrease training loss and increase the accuracy in model?

**URL:** <https://discuss.ai.google.dev/t/how-to-decrease-training-loss-and-increase-the-accuracy-in-model/31429>\
**Category:** General Discussion\
**Tags:** models, help\_request\
**Created:** [April 6, 2022, 9:28am UTC](https://discuss.ai.google.dev/t/how-to-decrease-training-loss-and-increase-the-accuracy-in-model/31429 "2022-04-06T09:28:55Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Chahine\_Mami](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/chahine_mami/32/10246_2.png) [@Chahine\_Mami](https://discuss.ai.google.dev/u/Chahine_Mami)\
**Post date:** [April 6, 2022, 9:28am UTC](https://discuss.ai.google.dev/t/how-to-decrease-training-loss-and-increase-the-accuracy-in-model/31429/1 "2022-04-06T09:28:55Z")

</div>

Hello  
I’m making a custom deep learning model. I’ve got a low accuracy value. My project is detecting scratches in optical fiber centrals. These images below explain the problem:

 ![DSCN0042](https://d3qe71uytubmmx.cloudfront.net/original/2X/4/42b3fe9ba2e17ca7965c153174a904a5fdb8ad0c.jpeg)  
 ![DSCN01 (10)](https://d3qe71uytubmmx.cloudfront.net/original/2X/d/d5414b3e1fd11bfb901e2df84c82c4ce8a639585.jpeg)  
 ![DSCN003 (16)](https://d3qe71uytubmmx.cloudfront.net/original/2X/d/d3ff34e00b64d637fdb24d536a8158c0715f7738.jpeg)  
I’ve made a custom dataset contain 200 images with good and bad images.  
I’ve tried a lot of models found in the net and it didn’t work well.  
I need to know the number of filters and value needed to this type of project.  
Thanks.

---

<div class="post-metadata">

**Author:** ![Kiran\_Sai\_Ramineni](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/kiran_sai_ramineni/32/8859_2.png) [@Kiran\_Sai\_Ramineni](https://discuss.ai.google.dev/u/Kiran_Sai_Ramineni)\
**Post date:** [April 6, 2022, 11:01am UTC](https://discuss.ai.google.dev/t/how-to-decrease-training-loss-and-increase-the-accuracy-in-model/31429/3 "2022-04-06T11:01:25Z")

</div>

Hi @Chahine_Mami due to less number of images in your dataset your model is getting low accuracy. you can improve your dataset size by performing [data augmentation](https://www.tensorflow.org/tutorials/images/data_augmentation) techniques on your dataset. Thanks!