-
Is possible to create a `prompt_process.general_mask(mask)`, in which `mask` can be an initial mask and not only points or bounding box(es)?
Currently, the `prompt_process` only supports points an…
-
what can i do to change the background color to black with the detected object?like the example as follow
![Snipaste_2023-10-19_22-53-58](https://github.com/CASIA-IVA-Lab/FastSAM/assets/67264710/9531…
-
In SAM, when utilizing YOLOv8, it is possible to segment a specific desired class from the COCO class dataset, rather than segmenting everything. However, when working with FASTSAM, I encountered a ch…
-
# FastSAM
* [arxiv](https://arxiv.org/abs/2306.12156)
* 本文是一篇technical report
* 主要思想是用CNN(YOLOv8-seg)代替ViT encoder提取特征,只用SAM 1/50的SA-1B数据,实现了运行速度50x faster(相较于ViT/H)和相近的分割性能
![image](https://g…
-
I transform FastSAM to onnx and the brief description in netron shows like this:
![Screenshot from 2023-08-22 16-25-24](https://github.com/CASIA-IVA-Lab/FastSAM/assets/18437178/34030cff-6a92-44c0-bb2…
-
If possible, could you share the evaluation code for Object proposal generation and instance segmentation in the paper? I have been searching for it for a long time but haven't found it yet.
-
### 问题确认 Search before asking
- [X] 我已经查询[历史issue](https://github.com/PaddlePaddle/PaddleDetection/issues),没有发现相似的bug。I have searched the [issues](https://github.com/PaddlePaddle/PaddleDetection/issu…
-
Hi, thx for your awesome code.
I have built the engine file fast_sam_1024.plan successfully using your code after modifying some parameters such as the image size. Additionally, I can get the result …
fdap updated
5 months ago
-
-
Hello, i have a big image (2000x3000) that i want to segment, with an output of ~1000 masks it overloads the memory of the GPU. Could we sequentially detach the images into the CPU so that we output t…